Aus der Abteilung für Infektions- und Tropenmedizin Leiter: Prof. Dr. med. Thomas Löscher Medizinische Poliklinik Innenstadt Kommisarischer Leiter: Professor Dr. med. Martin Reincke Klinikum der Universität München Untersuchungen zu Diagnostik und Therapie des Buruli Ulkus mittels tropenadaptierter Labormethoden in endemischen Regionen: Etablierung eines diagnostischen Netzwerkes in Ghana vorgelegt von Dr. med. Gisela Bretzel 2009
123
Embed
Untersuchungen zur Diagnostik des Buruli-Ulkus mittels ... · Das Buruli Ulkus ist gegenwärtig die dritt-, in einigen westafrikanischen Ländern bereits die zweithäufigste mykobakterielle
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Aus der Abteilung für Infektions- und Tropenmedizin
Leiter: Prof. Dr. med. Thomas Löscher
Medizinische Poliklinik Innenstadt
Kommisarischer Leiter: Professor Dr. med. Martin Reincke
Klinikum der Universität München
Untersuchungen zu Diagnostik und Therapie des Buruli Ulkus mittels
tropenadaptierter Labormethoden in endemischen Regionen:
Etablierung eines diagnostischen Netzwerkes in Ghana
Abkürzungsverzeichnis Abb. Abbildung AITM Abteilung für Infektions- und Tropenmedizin BNITM Bernhard Nocht Institut für Tropenmedizin bp Base pairs (Basenpaare) BUD Buruli Ulcer Disease cm Zentimeter d diem (Tag) DAHW Deutsche Lepra- und Tuberkulosehilfe e.V. DNA Desoxyribonucelic acid (Desoxyribonukleinsäure) DRB-PCR Dry reagent based PCR (Trockenreagenz-PCR) DT Drug Treatment (antimykobakterielle Behandlung) E. coli Escherichia coli EQA External Quality Assurance (externe Qualitätssicherung) FNA Feinnadelaspirat g Gramm GBUI Global Buruli Ulcer Initiative HIV Human immunodeficiency virus (Hi-Virus) hsp Heat Shock Protein (Hitzeschock Protein) IS Insertionssequenz ITS Internal transcribed spacer kb Kilobasen KCCR Kumasi Centre for Collaborative Research in Tropical Medicine kg Kilogramm M. Mycobacterium Mb Megabasen mg Milligramm mm Millimeter PCR Polymerase chain reaction (Polymerase-Kettenreaktion) RNA Ribonucleic acid (Ribonukleinsäure) ROM Reduced range of motion (Bewegungseinschränkung) rpoB RNA Polymerase Beta Untereinheit rRNA Ribosomale RNA S Svedberg SOP Standard Operating Procedure (Standardarbeitsanweisung) SPR Slide positivity rate (Positivitätsrate mikroskopischer Präparate) ST Surgical Treatment (chirurgische Behandlung) ST+ Surgical Treatment plus antimykobakterielle Therapie TAG Technical Advisory Group on Buruli Ulcer u.a. Unter anderem VNTR Variable number tandem repeats WHA World Health Assembly WHO World Health Organisation (Weltgesundheitsorganisation) z.B. Zum Beispiel
5
Einleitung Historischer Überblick
Die Erstbeschreibung ausgedehnter, zur klinischen Diagnose eines Buruli Ulkus („Buruli
Ulcer Disease“, BUD) passender Hautulzera stammt von dem britischen, in Uganda tätigen
Arzt Sir Albert Cook aus dem Jahr 1897. Peter MacCallum und Kollegen beschrieben im Jahr
1948 sechs weitere Fälle aus der Region um Bairnsdale nahe Melbourne, Australien. Ihnen
gelang die Erstisolierung des Erregers, Mycobacterium ulcerans. Aufgrund der Herkunft der
ersten australischen Patienten ist die Erkrankung im südlichen Australien bis heute auch unter
dem Namen „Bairnsdale ulcer“ bekannt. In den sechziger Jahren des 20. Jahrhunderts wurde
ein gehäuftes Auftreten des Buruli Ulkus im Buruli County in Uganda (heute: Nakasongola
District) beobachtet. Die derzeit meist gebrauchte Bezeichnung „Buruli Ulkus“ hat hier ihren
Ursprung. Zu Beginn der achtziger Jahre des letzten Jahrhunderts wurden zunehmend neue
Fälle aus weiteren Ländern, mit Schwerpunkt in West Afrika beschrieben. Die
Weltgesundheitsorganisation (WHO) gründete als Reaktion auf die stetig steigenden
Fallzahlen 1998 die Global Buruli Ulcer Initiative (GBUI). Die World Health Assembly
(WHA) verabschiedete schließlich im Jahr 2004 eine Resolution mit dem Ziel der
Verbesserung von Überwachung und Kontrollmaßnahmen, sowie einer Intensivierung der
Forschung (50, 51)
Erreger
Mycobacterium ulcerans, ein grampositives, säurefestes, nicht-chromogenes, langsam
wachsendes Stäbchen (Gruppe III der Runyon Klassifizierung) aus der Familie der
Mycobacteriaceae, Gattung Mycobacterium, zählt zu den atypischen Mykobakterien (MOTT:
mycobacteria other than tuberculosis, synonym NTM: non tuberculous mycobacteria). M.
ulcerans kann bei Temperaturen zwischen 29 und 33 °C auf Löwenstein-Jensen Nährböden
kultiviert werden (47). Das 5.8 Mb große M. ulcerans Genom, bestehend aus zwei zirkulären
Replikons, einem 5632 kb Chromosom und einem Virulenzplasmid, weist eine
Sequenzhomologie von über 98% mit Mycobacterium marinum auf. Eine Abspaltung von M.
marinum durch lateralen Gentransfer ist daher anzunehmen. Phylogenetischen Analysen
zufolge entstanden im weiteren evolutionsbiolgischen Verlauf zwei Abstammungslinien. Die
„ancestral lineage“ (Isolate aus Asien [China, Japan], Südamerika, Mexiko) weist einen hohen
genetischen Übereinstimmungsgrad mit M. marinum auf, die phylogenetische Evolution der
„classical lineage“ (Isolate aus Afrika, Australien, Süd-Ost Asien) dagegen führte zu
6
umfangreicher Reorganisation und Reduktion des Genoms und ermöglichte M. ulcerans eine
Anpassung an neue ökologische Nischen (23, 41). Mittels molekularbiologischer
Typisierungsmethoden, beispielsweise der Analyse von „variable number tandem repeats“
(VNTR), können gegenwärtig 11 M. ulcerans Genotypen unterschieden werden, die sich
gemäß der geographischen Herkunft der untersuchten Isolate in derzeit vier Cluster aufteilen
Abb. 2b Abb. 2d Abb. 2f Abb. 2. Diagnostische Untersuchungsmaterialien und Laboruntersuchungen (2a Abstrich, 2b Feinnadelaspirat, 2c Punch-Biopsie, 2d operativ entnommenes Gewebe, 2 e Mikroskopisches Abstrichpräparat, 2f M. ulcerans Kultur auf Jöwenstein-Jensen ) Unterschiedliche Faktoren wie beispielsweise Alter der Läsion/Erkrankungsdauer, sowie
Qualität der Probenabnahme und Probentransport können die diagnostischen Sensitivitäten
entscheidend beeinflussen. Generell sollten folgende Aspekte beachtet werden: Die
Probenabnahme von nicht-ulzerativen Läsionen sollte aus dem Zentrum der Läsion, von
ulzerativen Läsionen dagegen unter dem unterminierten Rand, im Grenzbereich zwischen
nekrotischem und makroskopisch gesundem Gewebe erfolgen. Abstriche ulzerativer Läsionen
werden durch kreisförmiges Umfahren der Läsion tief unter dem unterminierten Rand
gewonnen. Da die PCR-Untersuchung von Abstrichen in den meisten Fällen bereits zur
Sicherung der Diagnose führt, ist eine routinemäßige Gewebeabnahme bei ulzerativen Fällen
nicht erforderlich, sondern sollte nur bei negativem Abstrich und begründetem klinischen
Verdacht durchgeführt werden (5). Für den Transport der Proben ins Labor empfiehlt sich die
Tabelle 1. Mikroskopie- und PCR- Ergebnisse aus erstem und zweitem Untersuchungsgang (Übereinstimmungsrate zwischen controller und Testlabor, Anteil falsch negativer und falsch positiver Ergebnisse).
Die Mikroskopieergebnisse aus dem ersten Untersuchungsgang des Testlabors ergaben eine
SPR von 35.9%. Nach dem wiederholten Lesen diskrepanter Präparate im Testlabor im
zweiten Untersuchungsgang verbesserten sich sowohl die Übereinstimmungsrate zwischen
beiden Laboratorien, als auch Sensitivität (von 72.9% im ersten Untersuchungsgang auf
95,8%) und Spezifität (von 89.9% im ersten Untersuchungsgang auf 99.4%).
Die PCR-Ergebnisse aus dem ersten Untersuchungsgang des Testlabors ergaben eine
Positivitätsrate von 66%. Nach der wiederholten Analyse diskrepanter Proben im Testlabor
im zweiten Untersuchungsgang verbesserten sich sowohl die Übereinstimmungsrate zwischen
beiden Laboratorien, als auch Sensitivität (von 91.8% im ersten Untersuchungsgang auf
95.3%) und Spezifität (von 80.9% im ersten Untersuchungsgang auf 97.9%).
Ursachen für fehlerhafte Ergebnisse lagen beispielsweise in der Untersuchung einer zu
geringen Anzahl von Gesichtsfeldern (falsch negative Mikroskopieergebnisse) oder in der
Überinterpretation der Geldokumentation (falsch positive PCR-Ergebnisse).
Die in dieser Studie vorgestellte externe Qualitätssicherung und die damit verbundenen
Trainingsmaßnahmen konnten somit die Qualität der Diagnostik im Trainingslabor deutlich
verbessern und werden im künftigen Routinebetrieb nach dem gleichen System kontinuierlich
weitergeführt. Die Kosten für PCR-Untersuchungen sowie die Arbeitsbelastung des
controllers machen jedoch eine Reduktion der Stichprobengröße erforderlich.
20
A stepwise approach to the laboratory diagnosis of Buruli ulcer disease.
Tabelle 1. Diagnostische Sensitivität Gewebe-Mikroskopie, zusätzlicher Sensitivitätsgewinn bei Kombination mit Gewebe-PCR für nicht-ulzerative Läsionen
Durch mikroskopische Gewebeuntersuchungen konnten 40.3% der Buruli-Verdachtsfälle mit
nicht-ulzerativen Läsionen bestätigt werden, die zusätzliche Untersuchung mikroskopisch
negativer Gewebepräparate mittels PCR ergab weitere 25.4% positive Labordiagnosen.
Insgesamt konnten 65.7% der untersuchten Fälle durch die Kombination beider Methoden
Tabelle 2: Diagnostische Sensitivität Abstrich-Mikroskopie, zusätzlicher Sensitivitätsgewinn bei Kombination mit Abstrich-PCR, Gewebe-Mikroskopie, und Gewebe-PCR
Durch mikroskopische Abstrichuntersuchungen konnten 29.8% der Buruli-Verdachtsfälle mit
ulzerativen Läsionen bestätigt werden, die zusätzliche Untersuchung mikroskopisch negativer
Abstrichpräparate mittels PCR ergab weitere 38.3% positive Labordiagnosen. Durch
23
mikroskopische Untersuchung von Gewebepräparaten wurden weitere 4.3%, durch Gewebe-
PCR weitere 2.1% der Verdachtsfälle bestätigt. Insgesamt konnten 74.5% der untersuchten
Fälle durch die Kombination von Abstrich- und Gewebeuntersuchung mittels Mikroskopie
und PCR bestätigt werden.
Der Kostenanalyse ergab eine signifikante Reduktion der Testkosten bei sukzessiver
Anwendung (nicht-ulzerativ: 1125 €, ulzerativ: 1250 €) im Vergleich zu simultaner
Anwendung (nicht-ulzerativ: 1725 €, ulzerativ: 3250 €) der genannten Labormethoden.
Die in dieser Studie erhobenen Daten erlauben folgende Schlussfolgerungen: Wird statt der
bisher geforderten zwei positiven Laborergebnisse einer positiven Diagnose nur ein positives
Laborresultat als Kriterium zugrunde gelegt, erhöht sich der Anteil positiver Diagnosen um
rund 20%. Im Hinblick auf die hohe Spezifität von Mikroskopie und PCR (in dieser Studie
96.6% bzw. 100%) ist ein zweiter Bestätigungstest nicht erforderlich. Die bisher geltenden
WHO-Empfehlungen sollten somit zugunsten eines höheren Anteils positiver Labordiagnosen
revidiert werden.
Die Labordiagnostik des Buruli Ulkus sollte im Sinne einer Stufendiagnostik wie folgt
durchgeführt werden: Mikroskopische Untersuchung von Gewebepräparaten nicht-ulzerativer
Läsionen erlaubt unseren Daten zufolge einen Erregernachweis in etwa 40% der klinisch
diagnostizierten Verdachtsfälle. Die anschließende PCR-Untersuchung mikroskopisch
negativer Proben ermöglicht weitere etwa 25% positive Diagnosen. In beiden
Untersuchungen negative Gewebeproben sollten einer histopathologischen Analyse
unterzogen werden. Da die Verarbeitung von Gewebepräparaten für die Mikroskopie unter
einer Sicherheitswerkbank erfolgen sollte, ist diese Untersuchung in der Regel nicht im Labor
des peripheren Krankenhauses durchführbar.
Aus ulzerativen Läsionen entnommene Abstriche sollten zunächst mikroskopisch (zu
erwartende Sensitivität etwa 30%), mikroskopisch negative Abstriche dann mittels PCR
(zusätzliche Sensitivität etwa 40%) untersucht werden. Fallen beide Abstrichuntersuchungen
negativ aus (in etwa 30% der Fälle), können Gewebepräparate, analog zum oben
beschriebenen Vorgehen bei nicht-ulzerativen Läsionen mittels Mikroskopie und PCR
untersucht werden. Da hierbei jedoch nur etwa 6% zusätzliche positive Diagnosen erzielt
werden können, ist vor allem aus Kostengründen von einer Gewebeuntersuchung mittels PCR
abzuraten. Bei negativem Ergebnis der Abtrichuntersuchung sollten Gewebepräparate wenn
möglich histologisch untersucht werden.
24
Im Vergleich zur simultanen Untersuchung diagnostischer Proben ermöglicht die
Stufendiagnostik des Buruli Ulkus eine erhebliche Reduktion der Testkosten um bis zu 35%
für nicht-ulzerative und bis zu 60% für ulzerative Läsionen.
25
Dry reagent-based polymerase chain reaction compared with other laboratory methods
available for the diagnosis of Buruli ulcer disease.
Siegmund V, Adjei O, Nitschke J, Thompson W, Klutse E, Herbinger KH, Thompson R, van
Tabelle 1: Übereinstimmungsraten zwischen DRB-PCR und konventioneller Standard PCR für Abstriche und Gewebe aufgeschlüsselt nach Probenherkunft aus Behandlungszentrum A, AG, D, und G.
Tabelle 2: Gesamt-Positivitätsraten DRB-PCR, Standard-PCR, Mikroskopie, Kultur und Histopathologie nach Probenart in % Aufgeschlüsselt nach der Herkunft der Proben aus den verschiedenen Behandlungszentren
ergaben sich wiederum unterschiedliche Positivitätsraten. So wurden beispielsweise für die
DRB-PCR Untersuchung von Abstrichen vergleichbar hohe Werte für Zentrum A, AG, und D
ermittelt, während ein deutlich geringerer Anteil von Gewebeproben aus Zentrum D positiv
getestet wurde (Tabelle 3).
27
Behandlungszentrum Positivitätsrate pro Probenart und Test (%)
Tabelle 3: Positivitätsraten aller angewandten Untersuchungsmethoden aufgeschlüsselt nach Art der diagnostischen Probe und Probenherkunft aus Behandlungszentrum A, AG, D, und G. Die histopathologische Untersuchung von 167 Gewebeproben konnte in 130 Fällen die
klinische BUD-Verdachtsdiagnose bestätigen, in 37 Fällen ausschließen. Für 27 dieser
Patienten konnte eine Differentialdiagnose (meist bakterielle oder parasitäre Erkrankungen)
gestellt werden.
Die während einer zweijährigen Erprobungsphase unter tropischen Bedingungen erhobenen
Erfahrungswerte und Daten zur Leistungsfähigkeit der DRB-PCR im Vergleich mit anderen
Tabelle 1 zeigt die Sensitivitäten der untersuchten diagnostischen Teste an Gewebeproben
von BUD-Patienten mit nicht-ulzerativen Läsionen, stratifiziert nach Behandlungsgruppen.
Behandlungs- gruppe
Diagnostische Probe
DRB-IS2404 PCR
Mikroskopie IS2404-PCR bestätigte Kultur
DT Punch-Biopsie 93.5% (86/92) 57.6% (53/92) 70.8% (51/72) ST Exzidiertes
Gewebe 66.7% (10/15) 40.0% (6/15) 40.0% (6/15)
ST+ Ezidiertes Gewebe
85.7% (6/7) 85.7% (6/7) 0.0% (0/7)
Gesamt Alle Proben 89.5% (102/118 57.0% (65/114) 60.6% (57/94) Tabelle 1: Sensitivitäten von PCR, Mikroskopie, und IS2404-PCR bestätigter-Kultur in Gewebeproben von BUD-Patienten mit nicht-ulzerativen Läsionen, stratifiziert nach Behandlungsgruppen Die Sensitivität der PCR lag insgesamt signifikant höher als die Sensitivitäten von
Mikroskopie und Kultur. Die Untersuchung von Punch-Biopsien in der Gruppe der DT-
Patienten ergab signifikant höhere Sensitivitäten von PCR und Kultur als die Untersuchung
chirurgisch exzidierter Gewebeproben von ST und ST+ Patienten.
Tabelle 2 zeigt die Sensitivitäten der untersuchten diagnostischen Teste an Abstrich- und
Gewebeproben von BUD-Patienten mit ulzerativen Läsionen, stratifiziert nach
Alle Proben 69.8% (169/242) 44.6% (108/242) 25.0% (52/208) Tabelle 2: Sensitivitäten von PCR, Mikroskopie, und IS2404-PCR bestätigter-Kultur an Abstrich- und Gewebeproben von BUD-Patienten mit ulzerativen Läsionen, stratifiziert nach Behandlungsgruppen. Die Sensitivität der PCR lag insgesamt signifikant höher als die Sensitivitäten von
Mikroskopie und Kultur. Die PCR-Untersuchung von Abstrichen ergab in allen
Behandlungsgruppen signifikant höhere Sensitivitäten als die Untersuchung von
Gewebeproben.
Wir konnten in dieser Studie keinen Zusammenhang zwischen Erkrankungsdauer und
Testsensitivität nachweisen. Die Dauer antimykobakterieller Therapie vor Probenentnahme
korreliert jedoch eindeutig mit der Sensitivität von Kultur und PCR. Antimykobakterielle
Therapie von mehr als 20 bzw. 40 Tagen führte in unserem untersuchten Kollektiv zu einem
signifikanten Abfall der diagnostischen Sensitivität von Kultur und PCR.
In der bislang größten, vergleichenden Studie zur Sensitivität gegenwärtig verfügbarer
diagnostischer Methoden nach Einführung antimykobakterieller Therapie in West Afrika
ergab sich für die IS2404 PCR in allen untersuchten Untergruppen unserer Studienkohorte die
höchste Sensitivität.
Für die Diagnostik von Patienten aus unterschiedlichen Behandlungsgruppen mit
unterschiedlichen Läsionstypen können aus dieser Studie folgende Schlussfolgerungen
gezogen werden: Aufgrund der hohen Sensitivität von PCR und Kultur können 3 mm Punch-
Biopsien uneingeschränkt zur Diagnostik nicht vorbehandelter, nicht-ulzerativer Läsionen
empfohlen werden. Die PCR-Untersuchung von 4 und 6 mm Punch-Biopsien (30) bietet
keinerlei diagnostischen Vorteile. Die PCR-Untersuchung von Abstrichen aus ulzerativen
Läsionen ergab in allen Behandlungsgruppen, insbesondere aber bei nicht vorbehandelten
Patienten mit frühen Läsionen, signifikant höhere Sensitivitäten als die Untersuchung von
Gewebeproben. Die Untersuchung von Abstrichen mittels PCR kann somit als Methode der
33
Wahl zur Diagnostik ulzerativer Läsionen gelten. Vernarbungsprozesse im Krankheitsverlauf
können allerdings die Abnahme von Abstrichen erschweren und die Untersuchung von
Gewebeproben erforderlich machen.
Die Dauer antimykobakterieller Therapie vor Probenabnahme beeinflusst die Sensitivitäten
von Kultur und PCR. Während die diagnostische Sensitivität der Kultur in den ersten
Behandlungswochen bis auf 6% zurückging, war M. ulcerans DNA jedoch auch nach mehr
als sechswöchiger Behandlung in 50% der vorbehandelten Patienten nachweisbar. Aufgrund
der Persistenz mykobakterieller DNA eignet sich somit nur die Kultur zur Therapiekontrolle.
In Übereinstimmung mit unseren bisherigen Daten zeigte diese Studie weiterhin, dass über
20% mehr Verdachtsdiagnosen bestätigt werden können, wenn nur ein positives Testergebnis
einer positiven Diagnose zugrunde gelegt wird (5). Wie in unseren früheren Untersuchungen
ergab die histopathologische Analyse von in allen anderen Laboruntersuchungen negativen
Gewebeproben auch in diesem Patientenkollektiv 30% an zusätzlichen BUD-Diagnosen (5,
36, 37). Die Positivitätsraten (Anzahl positiver Testergebnisse eines Testes bezogen auf die
Gesamtzahl der untersuchten BUD-Verdachtsfälle) sowie die Gesamt-Sensitivitäten der in
dieser Studie untersuchten diagnostischen Methoden waren anderen bisher durchgeführten
Untersuchungen vergleichbar (3, 14, 15, 26, 31, 36, 39, 55). die Sensitivität der PCR lag
insgesamt signifikant höher als die Sensitivitäten von Mikroskopie und IS2404-PCR-
bestätigter Kultur. Bei der Mehrzahl aller bewachsenen Kulturen handelte es sich um M.
ulcerans. Von einigen Studienpatienten wurden jedoch während oder nach
antimykobakterieller Therapie auch andere Mykobakterienspezies isoliert. Bei wiederholtem
kulturellem Mykobakterien-Nachweis von antibiotisch therapierten Patienten sollte in jedem
Fall die Identifikation der Kulturen mittels molekularer Methoden angestrebt werden, um Co-
oder Superinfektionen mit anderen Mykobakterienspezies nachzuweisen.
34
Post-surgical assessment of excised tissue from patients with Buruli ulcer disease:
progression of infection in macroscopically healthy tissue.
Bretzel G, Siegmund V, Racz P, van Vloten F, Ngos F, Thompson W, Biason P, Adjei O,
Fleischer B, Nitschke J . Trop Med Int Health 2005;10(11):1199-206.
Bis zur offiziellen Empfehlung der antimykobakteriellen Therapie des Buruli Ulkus im Jahr
2004 durch die WHO (52) galt die weiträumige chirurgische Exzision als Therapie der Wahl
(48). Therapieerfolg und Rezidivraten variierten je nach angewandter Operationstechnik (1, 9,
22, 44). Chirurgisches Vorgehen und Exzisionsgröße wurden allein von Augenmaß und
Erfahrung des Operateurs bestimmt. Der makroskopische Aspekt ist jedoch nicht beweisend
für eine Exzision im Gesunden. Zur Untersuchung der Ausbreitung der Infektion innerhalb
der Läsion und über die Exzisionsränder hinaus wurde in dieser Studie exzidiertes Gewebe
von laborbestätigten BUD-Patienten mittels verschiedener Labormethoden untersucht.
Zwanzig Patienten aus Kamerun (n = 10) und Ghana (n = 10) mit klinisch diagnostizierten
ulzerativen Buruli-Läsionen wurden in zwei Behandlungszentren (Hôpital de District, District
de Santé d’Akonolinga, Akonolinga, Kamerun; Agogo Presbyterian Hospital, Agogo, Ghana)
unter Allgemeinanästhesie operiert. Hierbei wurde die Läsion im makroskopisch gesunden
Gewebe exzidiert, wobei nekrotisches und subkutanes Fettgewebe vollständig und bis zur
Faszie entfernt wurden. Zur Laborbestätigung der klinischen Diagnose mittels IS2404 PCR,
Kultur, und Histopatholgie wurden aus dem exzidierten Gewebe diagnostische Proben aus
dem unter dem unterminierten Rand befindlichen Grenzbereich zwischen nekrotischem und
optisch gesundem Gewebe entnommen. Zur Bestimmung der Ausbreitung der Infektion
innerhalb der Läsion wurden nach genauer Vermessung des exzidierten Gewebes aus dem
zwischen Ulkusrand und äußerem Exzisionsrand liegenden Gewebe drei parallele (maximal
50 x 15 mm große) Gewebestreifen entnommen. Diese wurden wiederum in (maximal 10 x 5
mm große) Segmente unterteilt (im folgenden „Gradienten“ genannt). Zur Untersuchung der
Exzisionsränder mittels PCR wurde der gesamte verbleibende Exzisionsrand in (maximal 10
x 5 mm große) Segmente unterteilt. Aus anatomischen Gründen konnten nur 16 komplette
Exzisionsränder untersucht werden. Die Laboruntersuchung aller genannten Proben wurde
nach standardisierten Methoden im Kumasi Centre for Research in Tropical Medicine,
Kumasi, Ghana (KCCR), sowie im Bernhard Nocht Institut für Tropenmedizin, Hamburg
(BNITM) durchgeführt. Zur Quantifizierung der Bakterienlast wurden histopathologische
Befunde in die Kategorien „negative“ „mild“, „moderate“, und „marked“ unterteilt (14, 40,
47).
35
Die Untersuchung der insgesamt 83 Gradientensegmente konnte in allen 20 Läsionen M.
ulcerans DNA und/oder für eine Infektion charakteristische histopathologische
Veränderungen im gesamten untersuchten Gewebe vom inneren Ulkusrand bis zum äußeren
Exzisionsrand nachweisen. Lediglich für ein Gradientensegment (viertes von fünf Segmenten)
einer Läsion konnte kein positiver Nachweis erbracht werden - allerdings waren für das
äußerste Segment (Segment fünf) wiederum beide Nachweismethoden positiv.
Nach Untersuchung der Exzisionsränder wurden die 16 Läsionen nach dem Anteil PCR
positiver Randsegmente in drei Gruppen eingeteilt: 100% der Randsegmente positiv (n=10),
>50% der Randsegmente positiv (n =1), <50% der Randsegmente positiv (n=5).
Die in diversen Studien berichteten Unterschiede in Rezidiv- und Wundheilungsraten wurden
auf angewandte Anästhesie- und Operationstechniken, sowie die Erfahrung des Operateurs
zurückgeführt (1, 9, 22, 44). Wird nicht unter Allgemein- sondern unter Lokalanästhesie mit
primärem Wundverschluss operiert, könnte dies die Entstehung von Rezidiven begünstigen.
Unter Lokalanästhesie kann nicht weiträumig exzidiert werden, die primäre Wundnaht gilt
aus chirurgischer Sicht bei lokalen Infektionen als kontraindiziert, die Injektion von
Lokalanästhetika begünstigt zudem möglicherweise die Diffusion der Bakterien in die
Peripherie der Läsion.
Obwohl unsere Studienpatienten unter Allgemeinanästhesie operiert, und weite Exzisionen im
makroskopisch gesunden Gewebe durchgeführt wurden, konnten wir in allen Läsionen eine
Ausbreitung der Infektion bis zur Peripherie des exzidierten Gewebes einschließlich der
Exzisionsränder nachweisen. Die Bestimmung der Exzisionsgröße allein aufgrund des
makroskopisch gesunden Aspektes des die Läsion umgebenden Gewebes kann somit eine
Progression der Infektion in peripheres subkutanes Fettgewebe nicht sicher ausschließen.
Innerhalb eines Beobachtungszeitraumes von etwa zwei Monaten während des postoperativen
stationären Aufenthaltes trat bei keinem der Studienpatienten ein Rezidiv auf. Die klinische
Relevanz des Nachweises von M. ulcerans in Exzisionsrändern bezüglich des Auftretens von
Rezidiven ist jedoch nur durch Langzeitbeobachtung definitiv zu klären. Potentielle
Risikopatienten könnten durch PCR-Untersuchung von Gewebeproben aus Exzisionsrändern
identifiziert werden. Patienten mit positiven Befunden sollten in jedem Fall während eines
Beobachtungszeitraumes von mindestens einem Jahr regelmäßig auf das Auftreten von
Rezidiven untersucht werden. Die Kombination mit antimykobakterieller Therapie erscheint
zur Verhinderung möglicher Rezidive ratsam.
36
Excision of pre-ulcerative forms of Buruli Ulcer Disease: a curative treatment?
Herbinger KH, Brieske D, Nitschke J, Siegmund V, Thompson W, Klutse E, Awua-Boteng N
Y, Bruhl E, Kunaa L, Schunk M, Adjei O, Loescher T, Bretzel G. Infection 2008. Dec. 3 [Epub
ahead of print]
Seit 2004 gilt die antimykobakterielle Kombinationstherapie mit Rifampicin und
Streptomycin, je nach klinischem Verlauf alleine oder in Kombination mit chirurgischer
Exzision, als Therapie der Wahl für nicht-ulzerative und ulzerative Formen der Buruli-
Erkrankung (52). Aufgrund der weiträumigen Ausbreitung der Bakterien in makroskopisch
gesundes, die Läsion umgebendes Gewebe bei ulzerativen Formen kann die chirurgische
Exzision nicht mit hinreichender Sicherheit die Entfernung des gesamten infizierten Gewebes
gewährleisten (6, 33). Selbst wenige verbleibende Bakterien stellen ein Rezidivrisiko dar (34),
somit ist eine antimykobakterielle Therapie ulzerativer Läsionen in jedem Fall sinnvoll. In
nicht-ulzerativen Erkrankungsformen dagegen befinden sich die Erreger im Zentrum der
Läsion (33). Eine Exzision mit ausreichend breiten Exzisionsrändern sollte daher die
vollständige Entfernung des erregerhaltigen Gewebes ermöglichen. Erfahrungsgemäß führt -
im Gegensatz zur chirurgischen Therapie ulzerativer Läsionen – die chirurgische Entfernung
nicht-ulzerativer Läsionen nur sehr selten zu funktionalen Einschränkungen betroffener
Gelenke und Extremitäten. Weiterhin ist die stationäre Verweildauer der Patienten nach
Dry-Reagent-Based PCR as a Novel Tool for Laboratory Confirmationof Clinically Diagnosed Mycobacterium ulcerans-Associated Disease in
Areas in the Tropics Where M. ulcerans Is EndemicV. Siegmund,1* O. Adjei,2 P. Racz,1 C. Berberich,2 E. Klutse,3 F. van Vloten,1 T. Kruppa,2
B. Fleischer,1 and G. Bretzel1
Department of Medical Microbiology and Immunology, Bernhard Nocht Institute for Tropical Medicine, Hamburg, Germany,1
and Kumasi Centre for Collaborative Research in Tropical Medicine, Kwame Nkrumah University of Science andTechnology, Kumasi,2 and Dunkwa Government Hospital, Dunkwa-on-Offin,3 Ghana
Received 22 June 2004/Returned for modification 20 August 2004/Accepted 21 September 2004
After tuberculosis and leprosy, Buruli ulcer (BU), caused by Mycobacterium ulcerans, is the third mostcommon mycobacterial disease in immunocompetent humans. The disease occurs in tropical countries, withfoci in West Africa, Central Africa, and the western Pacific. BU is defined as an infectious disease involving theskin and the subcutaneous adipose tissue characterized by a painless nodule, papule, plaque, or edema,evolving into a painless ulcer with undermined edges and often leading to invalidating sequelae. Due to thefundamental lack of understanding of modes of transmission, disease control in endemic countries is limitedto early case detection through improved active surveillance and surgical treatment. The laboratory confir-mation of BU is complicated by the absence of a diagnostic “gold standard.” Therefore, misclassification anddelayed diagnosis of BU may occur frequently, causing a considerable socioeconomic impact in terms oftreatment costs due to prolonged hospitalization. In order to respond to the urgent need to develop reliabletools for early case detection and to overcome technical difficulties accompanying the implementation ofdiagnostic PCR procedures in tropical countries, a dry-reagent-based PCR formulation for the detection of M.ulcerans in diagnostic specimens has been developed at the Bernhard Nocht Institute for Tropical Medicine.Following technical and clinical validation, the assay has been successfully installed and field tested at theKumasi Centre for Collaborative Research in Tropical Medicine, Kumasi, Ghana. Preliminary results show anexcellent diagnostic sensitivity of >95%.
After tuberculosis and leprosy, Buruli ulcer (BU) is the thirdmost common mycobacterial disease in immunocompetent hu-mans. The incidence of BU has been on the rise worldwide.The disease occurs in tropical countries, with foci in WestAfrica, Central Africa, and the western Pacific. The diseasemainly affects impoverished inhabitants of remote rural areas,with a preference for children under the age of 15, and isbelieved to be associated with tropical and subtropical wet-lands. Epidemiological studies suggest that swamps and slowlyflowing water are the sources of the organism. Recent evidencesupports the view that waterborne insects may be involved inthe transmission of the infection. However, the precise sourceof infection is still unknown. The incidence and prevalence ofBU worldwide are not precisely defined, as adequate surveil-lance data based on accurate case confirmation data are lack-ing. However, prevalences for the disease up to �20% havebeen reported from various foci in countries where the diseaseis highly endemic. In Ghana in 1999, for example, the overallcrude national prevalence rate of active lesions was 20.7 per100,000, but the rate was 150.8 per 100,000 in the district wherethe disease was most endemic (1, 3, 6, 7, 11).
According to World Health Organization standard case def-initions, BU is defined as an infectious disease involving the
skin and the subcutaneous adipose tissue characterized by apainless nodule, papule, plaque, or edema and evolving into apainless ulcer with undermined edges. The lesion may lead toextensive scarring, contractures, and deformations with possi-ble total loss of articulation function. If left untreated, it mayeven result in loss of limbs or blindness.
Cases that meet these clinical definitions are consideredprobable cases. Confirmed cases require a positive laboratorydiagnosis. Because BU is associated with nonspecific clinicalmanifestations and an indolent course, every ulcer or nodule inan area of endemicity should be suspected as a Mycobacteriumulcerans infection until proven otherwise. Observations madeby various researchers suggest that misclassification of clini-cally diagnosed BU cases may occur frequently. Early andhealed lesions, especially, may be confused with other skindiseases endemic in tropical areas. The differential diagnosiscomprises infectious (e.g., abscess, onchoceriasis, leprosy, ele-phantiasis, yaws, scrofuloderma, mycosis, actinomycosis, her-pes, cutaneous leishmaniasis, tropical phagedenic ulcer, ve-nous ulcer, and noma) and other (e.g., insect bites; psoriasis;enlarged lymph nodes; lipoma and other neoplasms; vascular,diabetic, and varicose ulcers; and burns) conditions and posesdifficulties, especially in tropical settings with limited access tolaboratory facilities (8, 12).
Treatment of BU with antibiotics has been widely unsuccess-ful. Experimental drug treatment trials are under way. Thepresent standard of treatment is surgical removal of the af-fected tissue, eventually followed by skin grafting. Due to the
* Corresponding author. Mailing address: Department of MedicalMicrobiology and Immunology, Bernhard Nocht Institute for TropicalMedicine, Hamburg, Germany. Phone: 49 40 42 81 84 67. Fax: 49 40 4281 84 00. E-mail: [email protected].
271
fundamental lack of understanding of modes of transmission,effective prevention strategies have not yet been developed. Atpresent, disease control in countries of endemicity is limited toearly detection through improved active surveillance and sur-gical treatment to prevent the development of severe ulcer-ation and resulting disabilities (3, 6, 15).
The laboratory confirmation of BU is complicated by theabsence of a diagnostic “gold standard” and a resulting lack ofclarity regarding the sensitivities and specificities of differentdiagnostic laboratory assays. A positive laboratory diagnosis ofM. ulcerans-associated disease requires any two of the follow-ing findings: detection of acid-fast bacilli in a Ziehl-Neelsen-stained smear, positive culture of M. ulcerans (swab or tissuespecimen, confirmed by biochemical tests or IS2404 PCR),histopathological confirmation, and positive IS2404 PCR(swab and/or tissue specimen) (6, 14, 15). Whereas the sensi-tivities of smear microscopy and culture are relatively low,histopathology and different PCR assays targeting differentregions in M. ulcerans IS2404 and 16S rRNA genes providediagnostic sensitivities of �90%. PCR can serve as a highlysensitive and reliable tool for presurgical early diagnosis andpostsurgical laboratory confirmation of probable cases. Thus,PCR can prevent misclassification and subsequent inadequatetreatment of patients in countries of endemicity. In addition,PCR provides reliable laboratory-confirmed incidence andprevalence data (5, 9, 10, 13, 16).
However, conventional PCR techniques in most cases arenot available in countries of endemicity. In order to respond tothe urgent need to develop reliable tools for early case detec-tion and to overcome technical difficulties accompanying theimplementation of conventional diagnostic PCR procedures intropical countries, a dry-reagent-based PCR formulation forthe detection of M. ulcerans in diagnostic specimens has beendeveloped at the Bernhard Nocht Institute for Tropical Med-icine (BNITM). The method is based on the standard diagnos-tic IS2404 PCR developed by Stinear et al. (14), and due to theuse of lyophilized reagents, is well adapted to tropical condi-tions and convenient for application in the field. After techni-cal and clinical validation, the method has been successfullyinstalled at the Kumasi Centre for Collaborative Research inTropical Medicine (KCCR), Kumasi, Ghana.
This pilot study presents the first validation data obtainedunder tropical conditions from March 2002 until September2002.
MATERIALS AND METHODS
Ethical considerations. Ethical clearance for the study was obtained throughthe Ethics Committee of the School of Medical Sciences, Kumasi, Ghana. Verbalconsent was obtained from the study participants and heads of households orguardians.
Mycobacterial strains and DNA standard. The mycobacterial strains used inthis study (Table 1) were kindly provided by Francoise Portaels, Institute forTropical Medicine, Antwerp, Belgium; David Dawson, Queensland Diagnosticand Reference Center for Mycobacterial Diseases, Brisbane, Australia; OheneAdjei, Kumasi Center for Collaborative Research in Tropical Medicine, Kumasi,Ghana; and Sabine-Rusch Gerdes, German National Reference Center for My-cobacteria, Borstel, Germany.
M. ulcerans and Myvobacterium marinum strains were cultured on Loewen-stein-Jensen medium at 32°C; all other mycobacterial strains were cultured at37°C.
Mycobacterium leprae DNA was kindly provided by Paul Klatser, Institute forTropical Medicine, Amsterdam, The Netherlands.
Diagnostic specimens. The diagnostic specimens used in this study (swabs, n �41; tissue specimens, n � 46) were obtained from 48 clinically diagnosed patientswith M. ulcerans-associated disease treated at Dunkwa Governmental Hospital,Dunkwa-on-Offin, Upper Denkyira District, Ghana, from March 2002 until Sep-tember 2002.
Clinical and epidemiological information for each patient was taken from BU1
TABLE 1. Mycobacterial strains and genomic DNA used inthe study
Strain reference no. Species Geographicalorigin Sourcea
ITM 97-610 M. ulcerans Ghana ITMITM 3129 M. ulcerans Zaire ITMITM 5114 M. ulcerans Mexico ITMITM 5147 M. ulcerans Australia ITMITM 8756 M. ulcerans Japan ITMITM 9146 M. ulcerans Benin ITMITM 94-511 M. ulcerans Ivory Coast ITMITM 98-912 M. ulcerans China ITMITM 97-680 M. ulcerans Togo ITMITM 96-657 M. ulcerans Angola ITMITM 94-1328 M. ulcerans Malaysia ITMITM 5156 M. ulcerans Papua ITMQDRLMD 9807 M. ulcerans Australia QDRLMDQDRLMD 9808 M. ulcerans Australia QDRLMDQDRLMD 9819 M. ulcerans Australia QDRLMDQDRLMD 9820 M. ulcerans Australia QDRLMDQDRLMD 9885 M. ulcerans Australia QDRLMDQDRLMD 9920 M. ulcerans Australia QDRLMDQDRLMD 10128 M. ulcerans Australia QDRLMDQDRLMD 10137 M. ulcerans Australia QDRLMDQDRLMD 10166 M. ulcerans Australia QDRLMDQDRLMD 10463 M. ulcerans Australia QDRLMDKCCR 207 M. ulcerans Ghana KCCRKCCR 216 M. ulcerans Ghana KCCRKCCR 221 M. ulcerans Ghana KCCRKCCR 4 M. ulcerans Ghana KCCRKCCR 05 M. ulcerans Ghana KCCRKCCR 05 R/S M. ulcerans Ghana KCCRKCCR 07 M. ulcerans Ghana KCCRKCCR 7 M. ulcerans Ghana KCCRKCCR 10 M. ulcerans Ghana KCCRKCCR 11 M. ulcerans Ghana KCCRKCCR 12 M. ulcerans Ghana KCCRKCCR 13 M. ulcerans Ghana KCCRKCCR 14 M. ulcerans Ghana KCCRKCCR 19 M. ulcerans Ghana KCCRKCCR 21 M. ulcerans Ghana KCCRKCCR D1 M. ulcerans Ghana KCCRKCCR D3 M. ulcerans Ghana KCCR3240/02; BNITM/M1 M. marinum Germany Borstel10368/01; BNITM/M2 M. lentiflavium Germany Borstel9679/00; BNITM/M3 M. tuberculosis
H37RvGermany Borstel
3746/02; BNITM/M4 M. avium Germany Borstel827/02; BNITM/M5 M. intracellulare Germany BorstelBNITM/M6 M. fortuitum Germany BorstelBNITM/M7 M. szulgai Germany Borstel2968/02; BNITM/M8 M. xenopi Germany Borstel6554/01; BNITM/M9 M. scrofulaceum Germany Borstel3899/02; BNITM/M10 M. gordonae Germany Borstel3709/02; BNITM/M11 M. kansasii Germany Borstel2280/02; BNITM/M12 M. malmoense Germany BorstelBNITM/M13 M. chelonae Germany BorstelBNITM/M14 M. smegmatis Germany BorstelGenomic DNA M. leprae The Netherlands KIT
a ITM, Institute for Tropical Medicine, Antwerp, Belgium; QDRLMD,Queensland Diagnostic and Reference Laboratory for Mycobacterial Diseases;KIT, Institute for Tropical Medicine, Amsterdam, The Netherlands.
272 SIEGMUND ET AL. J. CLIN. MICROBIOL.
forms: ulcerative lesions, n � 37; nonulcerative lesions, n � 7 (nodules, n � 4;plaque, n � 1; papule, n � 1; edema, n � 1); clinical data not available, n � 4.
The duration of the disease ranged from 1 week to 16 years, with a medianduration of 2 months.
To standardize the process of specimen collection, the following criteria wereestablished. Diagnostic swabs (Greiner Bio-One, Essen, Germany) were to betaken from the undermined edges of the lesions before surgery; 10- by 10-mmtissue specimens were to be obtained from surgically excised tissue. In the caseof nodules, the tissue specimens should contain a section of the center of thenodule; in the case of ulcers, the specimens should be taken from the edge of theulcerative lesions containing necrotic tissue sections.
The PCR specimens were stored in 2-ml tubes (Sarstedt, Numbrecht, Ger-many) containing 700 �l of cell lysis solution (Puregene DNA isolation kit;Gentra Systems, Indianapolis, Ind.) at room temperature until they were pro-cessed.
PCR-negative tissue specimens were subjected to histopathological analysis toexclude other infectious and noninfectious conditions considered for differentialdiagnosis. Ten- by 10-mm tissue sections were stored in 5-ml tubes (Sarstedt) in10% formaldehyde, and histopathological analysis was carried out according tostandardized criteria (4).
DNA preparation. The Puregene DNA isolation kit was used for DNA isola-tion with minor modifications evaluated for extracting M. ulcerans DNA fromtissue samples as described below. All materials mentioned were included in thekit.
Tissue specimens were inactivated at 95°C for 15 min. Subsequently, thespecimens were cut, and 5-mm3 (maximum) pieces were incubated overnight at55°C in 700 �l of cell lysis solution enriched with proteinase K (Sigma-Aldrich,Munich, Germany) to a final concentration of 300 �g/ml. The proteinase K wasinactivated at 95°C for 15 min. After the specimens were cooled to room tem-perature, egg white lysozyme was added to a final concentration of 250 �g/ml,and the specimens were incubated at 37°C for 1 h.
Processing of swab samples, as well as subsequent procedures of DNA extrac-tion for tissue specimens, were carried out according to the manufacturers’instructions. The DNA pellets were resuspended in 200 �l of DNA hydrationsolution, and extracts were stored at 4°C until they were further processed.
Selection of oligonucleotides and PCR conditions. (i) Standard referencemethod. Based on the findings in a recent publication (14), amplification of aspecific 492-bp-long DNA sequence in M. ulcerans IS2404 (GenBank accessionno. AF003002) by PCR was carried out from 2 �l of target DNA extract usingprimers MU5 (5�-AGC GAC CCC AGT GGA TTG GT) and MU6 (5�-CGGTGA TCA AGC GTT CAC GA). The reaction volume was 20 �l, containing 10mM Tris-HCl (pH 8.3), 50 mM KCl, 2.5 mM MgCl2, 200 �M (each) de-oxynucleoside triphosphate, and 1 U of Ampli-Taq Gold DNA polymerase. Allreagents except the oligonucleotides (TibMolbiol, Berlin, Germany) were pur-chased from Perkin-Elmer, Weiterstadt, Germany. DNA extracts of diagnosticspecimens were tested in 100 and 10�1 dilutions. Hot-start thermal cycling wasconducted in a Primus Thermocycler (MWG Biotech, Ebersberg, Germany)according to the following cycling profile: one initial activation step with poly-merase at 94°C for 10 min; 40 cycles, each consisting of denaturation at 94°C for10 s, primer annealing at 58°C for 10 s, and extension at 72°C for 30 s; and onefinal extension step at 72°C for 15 min. The amplification products were held at4°C until they were further processed and detected by agarose gel electrophoresis(1.5%) and ethidium bromide staining (1 �g/ml).
(ii) Dry-reagent-based PCR. PuReTaq Ready-To-Go PCR Beads (AmershamBiosciences, Freiburg, Germany) were used to carry out PCR amplification.When brought to a final volume of 25 �l, each reaction mixture contained �2.5U of PuReTaq DNA polymerase, 10 mM Tris-HCl [pH 9.0], 50 mM KCl, 1.5 mMMgCl2, 200 �M (each) deoxynucleoside triphosphate, and stabilizers, includingbovine serum albumin. The oligonucleotides (MU5 and MU6) were lyophilized(Hetovac; Nunc GmbH, Wiesbaden, Germany) for 20 min (1.150 rpm) in 200-�lreaction tubes (MWG Biotech) and applied to the reaction in a concentrationequal to that in the standard reference method. DNA extracts of diagnosticspecimens were tested in 100 and 10�1 dilutions.
Because the reaction conditions of the PuReTaq Ready-To-Go PCR beadTechnology were fixed as described above and therefore deviated from thestandard reference method, the amount of target DNA was extrapolated andadjusted to 2.5 �l due to the increased reaction volume.
Technical validation and optimization of PCR. (i) Quantified plasmid DNAstandard. The 492-bp target region of the standard reference method was am-plified, analyzed by gel electrophoresis, and prepared using the NucleoSpinExtract kit (Macherey Nagel, Duren, Germany) according to the manufacturer’sinstructions. Subsequently, the amplicon was directly cloned via the T/A stickyend using the AdvanTAge PCR cloning kit (Clontech, Heidelberg, Germany).
Plasmid DNA was prepared using the NucleoSpin Plasmid kit (MachereyNagel). Afterwards, plasmid concentrations were determined photometrically(260 nm). Because the target region is located within a multiple-copy elementand is therefore detectable �50 times per genome (13), the plasmid standardwith 1.5 genome equivalents represents �75 copies of the element.
(ii) Validation by testing diagnostic specimens in comparison to the IS2404reference method. Diagnostic specimens used for validation of the dry-reagent-based PCR method were obtained from 48 clinically diagnosed BU patients. (i)Swabs (n � 19) and tissue specimens (n � 16) from 19 clinically diagnosed BUpatients were simultaneously tested at the BNITM by both the standard refer-ence and the dry-reagent-based PCR methods. (ii) After implementation of themethod at the KCCR, swabs (n � 22) and tissue specimens (n � 30) from 30clinically diagnosed BU patients were tested simultaneously at the KCCR (dry-reagent-based PCR) and the BNITM (standard reference method).
Internal quality control. (i) Extraction control. To exclude false-positive PCRresults caused by contamination during DNA extraction, an extraction was per-formed without a specimen and used as an extraction control. The extractioncontrol was required to produce a negative result after PCR.
(ii) PCR run controls. To exclude false-positive and -negative PCR results, thefollowing control reactions were performed.
(a) Negative control reaction. The reaction mixture contained water instead oftemplate DNA, which should not produce a specific band after PCR.
(b) Positive control reaction. PCR was performed using plasmid DNA con-taining the specific 492-bp DNA sequence of M. ulcerans IS2404 as a template.
(c) Inhibition control reactions. For quality testing of the DNA extract, inhi-bition control PCRs were run in parallel with the tested samples. These controlsconsisted of half of the volume of DNA extract, with the IS2404 plasmid DNAas a competitor (1.5 genome equivalents). Possible inhibitory effects of sub-stances in the DNA extract were excluded if a clearly visible PCR amplicon of492 bp was present in the inhibition control reactions.
RESULTS
Analytic sensitivity. To establish a dry-reagent-based PCRfor the detection of M. ulcerans, IS2404 was selected as thetarget region. This target had been proven to be appropriatefor diagnostic application, and its use is regarded as the stan-dard reference method routinely used for diagnostic purposes(14).
To determine the analytical sensitivity in a quantified DNAstandard, we cloned the amplicon of the standard referencemethod into Escherichia coli.
Replicates of log10 plasmid serial dilutions were amplified byboth the standard reference PCR and the dry-reagent-basedPCR. It was shown that the sensitivity was excellent: 1.5 ge-nome equivalents per reaction could be detected with the stan-dard reference PCR in 10 of 10 reactions. When testing arith-metically higher dilutions (�0.15 plasmids per reaction),positive standard reference PCR results were occasionally ob-tained due to the plasmids randomly distributed among thesamples. Therefore, the reaction conditions could be consid-ered already optimized.
The same sensitivity, i.e., 1.5 genome equivalents per reac-tion, was achieved using the dry-reagent-based PCR formula-tion, indicating comparable analytical sensitivity.
Sensitivity and specificity. (i) Mycobacterial referencestrains. To confirm the specificity of the dry-reagent-basedPCR, 39 M. ulcerans reference strains and 15 different myco-bacterial strains (Table 1) were tested with our assay. GenomicDNA preparations were obtained by extraction of culture ma-terial. All 39 M. ulcerans reference strains tested positive andall 15 other mycobacterial strains tested negative in the dry-reagent-based PCR.
(ii) Dry-reagent-based PCR in comparison to the IS2404reference method. Nineteen swabs and 16 tissue specimens
VOL. 43, 2005 DRY-REAGENT PCR CONFIRMATION OF M. ULCERANS-ASSOCIATED DISEASE 273
from 19 patients were tested simultaneously with both methodsat the BNITM. The accordance (swabs, 94.7%; tissue speci-mens, 75%) and discordance (swabs, 5.3%; tissue specimens,25.0%) rates are presented in Table 2.
After implementation of the dry-reagent-based PCR at theKCCR, 22 swab specimens and 30 tissue specimens from 30patients were tested simultaneously with both methods(KCCR, dry-reagent-based PCR; BNITM, standard referencemethod). The accordance (swabs, 95.5%; tissue specimens,96.7%) and discordance (swabs, 4.5%; tissue specimens, 3.3%)rates are presented in Table 3.
(iii) Diagnostic sensitivity. Totals of 31.6 (6 of 19) and27.3% (6 of 22) of the swabs tested positive, and 18.8 (3 of 16)and 36.7% (11 of 30) of the tissue specimens produced apositive PCR result; 63.2 (12 of 19) and 68.2% (15 of 22) of theswabs tested negative, and 56.3 (9 of 16) and 60% (18 of 30) ofthe tissue specimens, i.e., 58.7% of the total number of tissuespecimens (27 of 46), produced a negative PCR result (Tables2 and 3).
Histopathological analysis of 27 PCR-negative tissue spec-imens. Eleven specimens were inappropriate for diagnosticpurposes, as the subcutaneous adipose tissue was either miss-ing or not complete. For two specimens, histopathological re-sults were not available. Unspecific dermatitis was found in twospecimens, and one specimen was diagnosed as Kaposi’s sar-coma. Two specimens showed signs of parasitic infection, oneof those being onchocerciasis. One specimen was confirmed ascutaneous tuberculosis.
The histopathological features of eight specimens were com-patible with M. ulcerans-associated disease. Six of those couldbe classified as active disease, one could be classified as inac-
tive (healing) stage, and one case could not be definitely con-firmed.
DISCUSSION
At present, control strategies for BU in countries of ende-micity are limited to early case detection through improvedactive surveillance and surgical treatment. A diagnostic goldstandard for the laboratory confirmation of BU has not yetbeen established, and sensitive diagnostic techniques like PCRand histopathology are often not available in areas of ende-micity (6). Thus, misclassification and delayed diagnosis mayoccur frequently. According to a retrospective study carriedout in Ashanti Region, Ghana, in 1994 to 1996, the averagetotal treatment costs for a BU patient with advanced ulcerativedisease requiring prolonged hospitalization were determinedto be U.S.$780 as opposed to U.S.$20 to $30 for early cases.Thus, early case detection and subsequent surgical treatmentreduce patient-related treatment costs (2).
In order to respond to the urgent need to develop reliabletools for early case detection, a dry-reagent-based PCR for-mulation for the detection of M. ulcerans in diagnostic speci-mens has been developed at the BNITM.
The implementation of conventional PCR assays in tropicalcountries is accompanied by various technical difficulties. Thetransport and storage of reagents and specimens require coldchains and equipment like generators and voltage stabilizers toprevent damage to the reagents and specimens by repeatedfreezing and thawing due to regularly occurring power cuts.Furthermore, conventional PCR assays require careful han-
TABLE 2. Accordance and discordance rates (BNITM testing)a
Specimen
Result
% %accordance
Result
% %discordanceDry-reagent-based
PCRStandard reference
methodDry-reagent-based
PCRStandard reference
method
Swab Positive Positive 31.6 (6 of 19) Positive Negative 5.3 (1 of 19)Negative Negative 63.2 (12 of 19) Negative Positive 0 (0 of 19)
Total 94.7 5.3
Tissue Positive Positive 18.8 (3 of 16) Positive Negative 6.3 (1 of 16)Negative Negative 56.3 (9 of 16) Negative Positive 18.8 (3 of 16)
Total 75.0 25.0
a Dry-reagent-based PCR (BNITM)/standard reference method (BNITM).
TABLE 3. Accordance and discordance rates (KCCR and BNITM testing)a
Specimen
Result
% %Accordance
Result
% %DiscordanceDry-reagent-based
PCRStandard reference
methodDry-reagent-based
PCRStandard reference
method
Swab Positive Positive 27.3 (6 of 22) Positive Negative 4.5 (1 of 22)Negative Negative 68.2 (15 of 22) Negative Positive 0 (0 of 22)
Total 95.5 4.5
Tissue Positive Positive 36.7 (11 of 30) Positive Negative 3.3 (1 of 30)Negative Negative 60.0 (18 of 30) Negative Positive 0.0 (0 of 30)
Total 96.7 3.3
a Dry-reagent-based PCR (KCCR)/standard reference method (BNITM).
274 SIEGMUND ET AL. J. CLIN. MICROBIOL.
dling of reaction components and strict measures by skilledlaboratory workers to avoid contamination of reagents.
The use of lyophilized, room-temperature-stable PCR re-agents and transport buffer for specimens prevents tempera-ture-dependent transport- and storage-related problems. Inaddition, lyophilized primers and reaction mixtures are easy tohandle and less time-consuming for laboratory staff. The chem-icals are not sensitive to climatic conditions, like heat andhumidity, and thus, the qualities of reagents and reactions arealways equal. The risk of contamination is minimized, becauseinstead of at least eight different liquid components, as inconventional PCR assays, only water and template are addedto the lyophilized reagents.
The dry-reagent-based PCR is slightly more expensive thanthe conventional diagnostic PCR (in the range of €2 to €3,depending on the manufacturer); however, the numerous ad-vantages of the assay clearly outweigh the slight financial im-balance. In general, studies of the cost-effectiveness of imple-menting highly sensitive diagnostic tools like PCR in areas ofhigh endemicity in terms of reducing total treatment costs arerequired.
With an analytical sensitivity and specificity equal to those ofconventional PCR and accordance rates for diagnostic swabs (n �19) of 95% (Table 2) for both dry-reagent-based and standardreference methods carried out simultaneously at the BNITM, weconsidered the dry-reagent-based PCR a reliable diagnostic tool.As the discordance rate of 25% (Table 2) for tissue specimens(n � 16) was attributable to technical aspects of specimen collec-tion, i.e., the specimens subjected to the two methods were notobtained from the same location, there was no evidence for alower sensitivity of the method for tissue samples.
After implementation of the assay at the KCCR, 52 diag-nostic specimens from 30 clinically diagnosed patients weresimultaneously tested at the KCCR (dry-reagent-based assay)and BNITM (standard reference method). Accordance ratesof �95% (Table 3) for both swabs and tissue specimens sug-gest that the dry-reagent-based PCR assay is highly reliableand well adapted to application under tropical conditions.
The diagnostic specimens tested for technical validation atthe BNITM and for validation under tropical conditions simul-taneously at the KCCR and the BNITM show relatively lowdiagnostic sensitivity (31.6 and 27.3%, respectively, for swabs;18.8 and 36.7%, respectively, for tissue samples [Tables 2 and3]). As the aim of this study was to establish and validate thedry-reagent-based diagnostic PCR by comparing it with thestandard IS2404 reference method, all patients with clinicallydiagnosed M. ulcerans-associated disease who presented atDunkwa Government Hospital during the study period wereincluded in the pilot study. Therefore, specific selection crite-ria, like the type and size of lesion or duration of the disease,were not applied. Preliminary data from an ongoing study atthe KCCR aimed at the PCR confirmation of clinically diag-nosed early cases suggest high diagnostic sensitivity (�90%)for early lesions (i.e., where the duration of disease was �6months). Of the patients included in the validation pilot study,however, 47.9% had lesions older than 6 months (13 of 48;27.1%) or data were not available (10 of 48; 20.8%). In thosecases, the age of the lesions was likely to affect the diagnosticsensitivity, as older lesions might be in an inactive (healing)stage, and thus, bacilli could not be detected.
In order to explain the relatively high number of PCR-negative tissue specimens (27 of 46; 58.7%), they were sub-jected to histopathology. Histopathological analysis revealedthat 40.7% (11 of 27) of the specimens were not appropriatefor diagnostic purposes, as the subcutaneous tissue was eithermissing or incomplete. In 22.2% (6 of 27) of the specimens,either noninfectious (nonspecific dermatitis and Kaposi’s sar-coma) or other infectious (tuberculosis, onchocerciasis, andother parasitic infections) conditions were diagnosed. Of thePCR-negative tissue specimens, 29.6% (8 of 27) showed his-topathological features compatible with Buruli ulcer but with-out detectable acid-fast bacilli. One specimen was in the inac-tive (healing) stage. Older and inactive lesions tend not tocontain detectable amounts of bacilli and thus may not pro-duce positive PCR results. Therefore, only early lesions mightbe considered suitable for molecular diagnosis of M. ulcerans-associated disease. Furthermore, these findings clearly supportthe need to establish the differential diagnosis before classify-ing and treating a patient for Buruli ulcer. Thus, diagnostictools to exclude other conditions that might be mistaken for M.ulcerans-associated disease are required as well. These findingsalso stress the importance of accurate specimen collection, asonly specimens that are taken from necrotic-tissue areas andthat include the subcutaneous tissue allow a reliable diagnosisof M. ulcerans-associated disease. Despite the prior definitionof criteria for specimen collection, in this pilot study, the tech-nical procedures were not closely monitored, as a close linkbetween the hospital and the laboratory was not yet fully es-tablished. The diagnostic sensitivity clearly depends on thequality of specimens, and only a close collaboration betweenthe surgeon and the laboratory guarantees optimized labora-tory results.
If these conditions are fulfilled, the dry-reagent-based PCRpresented in this study can serve as a reliable and rapid tool forthe laboratory confirmation of M. ulcerans-associated diseaseunder tropical conditions. Implementation of the method inmajor treatment centers in areas of endemicity is under wayand will provide diagnostics at the district level.
ACKNOWLEDGMENTS
We are grateful to Christian Drosten (BNITM) for stimulating dis-cussion and technical advice. We thank Serwaa Akosua (KCCR), Sab-ine Kohler, Monika Picker, Petra Plahn, Julia Polansky, Anja Schorle,Petra Meyer, and Birgit Raschdorff (BNITM) for excellent technicalassistance. We thank Jorg Nitschke for providing his surgical expertise.We are especially grateful to Christel Schmetz (BNITM) for her con-tinuous support.
REFERENCES
1. Amofah, G., F. Bonsu, C. Tetteh, J. Okrah, K. Asamoa, K. Asiedu, and J.Addy. 2002. Buruli ulcer in Ghana: results of a national case search. Emerg.Infect. Dis. 8:167–170.
2. Asiedu, K., and S. Etuaful. 1998. Socioeconomic implications of Buruli ulcerin Ghana: a three-year review. Am. J. Trop. Med. Hyg. 59:1015–1022.
3. Asiedu, K., M. Raviglione, and R. Scherpbier (ed.). 2000. Buruli ulcer.Mycobacterium ulcerans infection. World Health Organization, Geneva,Switzerland.
4. Guarner, J., J. Bartlett, E. A. Whitney, P. L. Raghunathan, Y. Stienstra, K.Asamoa, S. Etuaful, E. Klutse, E. Quarshie, T. S. van der Werf, W. T. A. vander Graaf, C. H. King, and D. A. Ashford. June 2003, posting date. His-topathologic features of Mycobacterium ulcerans infection. Emerg. Infect.Dis. [Online.] http:/www.cdc.gov/ncidod/EID/vol9no6/02–0485.htm.
5. Guimaraes-Peres, A., F. Portaels, P. de Rijk, K. Fisette, S. R. Pattyn, J. P.van Vooren, and P. A. Fonteyne. 1999. Comparison of two PCRs for detec-tion of Mycobacterium ulcerans. J. Clin. Microbiol. 37:2006–2008.
VOL. 43, 2005 DRY-REAGENT PCR CONFIRMATION OF M. ULCERANS-ASSOCIATED DISEASE 275
6. King, C. H., D. A. Ashford, K. Dobos, E. A. S. Whitney, P. L. Raghunathan,J. Guarner, and J. W. Tappero. 2001. Mycobacterium ulcerans infection andBuruli ulcer disease: emergence of a public health dilemma, p. 137–152. InW. M. Sheld, W. A. Craig, and J. M. Hughes (ed.), Emerging infections 5.ASM Press, Washington, D.C.
7. Meyers, W. M., N. Tignokpa., G. B. Priuli., and F. Portaels. 1996. Mycobac-terium ulcerans infection (Buruli ulcer): first reported patients in Togo. Br. J.Dermatol. 134:1116–1121.
8. Mwanatambe, M., M. Yajima, S. Etuaful, K. Suzuki, K. Nishigai, K. Asiedu,N. Yamada, and G. Asano. 2002. Malignant melanoma of the buttock pre-senting as a Buruli ulcer. J. Dermatol. 29:357–361.
9. N�Guessan, K., Y. Kouassi, S. Bouzid, P. Ehuie, K. Koffi, C. Oniangue, N.Aka, and M. Dosso. 2001. Value and limits of microscopy of exudates inMycobacterium ulcerans cutaneous infection in Cote d’Ivoire. Bull. Soc.Pathol. Exot. 94:9–10.
10. Portaels, F., J. Agular, K. Fisette, P. A. Fonteyne, H. De Beenhouver, P. deRijk, A. Guedenon, R. Lemans, C. Steunou, C. Zinsou, J. M. Dumonceau,and W. M. Meyers. 1997. Direct detection and identification of Mycobacte-rium ulcerans in clinical specimens by PCR and oligonucleotide specificcapture plate hybridizytion. J. Clin. Microbiol. 35:1097–1100.
11. Portaels, F., P. Elsen, A. Guimares-Peres, P. Fonteyne, and W. M. Meyers.
1999. Insects in the transmission of Mycobacterium ulcerans infection. Lancet353:986.
12. Portaels, F., P. Johnson, and W. M. Meyers (ed.). 2001. Buruli ulcer. Diag-nosis of Mycobacterium ulcerans disease. A manual for health care providers.World Health Organization, Geneva, Switzerland.
13. Ross, B. C., L. Marino, F. Oppedisano, R. Edwards, R. M. Robins-Browne,and P. D. Johnson. 1997. Development of a PCR assay for rapid diagnosis ofMycobacterium ulcerans infection. J. Clin. Microbiol. 35:1696–1700.
14. Stinear, T., B. C. Ross, J. K. Davies, L. Marino, R. M. Robins-Browne, F.Oppedisano, A. Sievers, and P. D. Johnson. 1999. Identification and char-acterization of IS2404 and IS2606: two distinct repeated sequences for de-tection of Mycobacterium ulcerans by PCR. J. Clin. Microbiol. 37:1018–1023.
15. Whitney, E. A. S., M. Phelan, P. L. Raghunathan, Y. Stienstra, K. Dobos, J.Guarner, S. Kilhstrom, A. Ablordey, S. Etuaful, E. Klutse, E. Quarshie, D.Ofori-Adjei, T. S. van der Werf, W. T. A. van der Graaf, K. Asamoah, G.Amofah, C. H. King, and D. A. Ashford. 2002. Latent class analysis (LCA)evaluation of four diagnostic tests for Buruli ulcer disease. Presented at the5th W.H.O. Advisory Group Meeting on Buruli Ulcer, Geneva, Switzerland.
16. World Health Organization. 2002. Weekly epidemiological record, vol. 77, p.271–272.
276 SIEGMUND ET AL. J. CLIN. MICROBIOL.
External quality assurance for the laboratory diagnosis of Buruli
ulcer disease in Ghana
G. Bretzel1,2, V. Siegmund1,2, J. Nitschke2, K. H. Herbinger1, R. Thompson2, E. Fleischmann1, B. Fleischer2 and
O. Adjei3
1 Department of Infectious Diseases and Tropical Medicine, University of Munich, Munich, Germany2 Bernhard Nocht Institute for Tropical Medicine, Hamburg, Germany3 Kumasi Centre for Collaborative Research in Tropical Medicine, Kumasi, Ghana
Summary objective To assure the quality of the laboratory diagnosis of Buruli ulcer disease; microscopy and
PCR were subjected to external quality assurance (EQA).
methods Slides were read by test laboratory staff, followed by blinded re-reading by the controller.
Parallel testing of PCR specimens was carried out at the local and external reference laboratory. Slides
and PCR specimens with discordant results were subjected to a second reading/testing by the controller
to determine the final result. For training purposes, slides and PCR specimens with discrepant results
were subsequently re-read/re-tested under supervision at the test laboratory.
Table 4 shows the results of the test laboratory relating
to the final results of the controller after the second reading
of slides with discrepant results at the test laboratory
relating to the total number of slides. The percentage of
false negatives dropped from 27.1% to 4.2% (5/118), with
3.4% high false negatives (4/118) and 0.8% low false
negatives (1/118). The percentage of high false positives fell
from 10.1% to 0.6% (1/169). The second reading resulted
in a concordance rate of 97.9% (281/287) and a discord-
ance rate of 2.1% (6/287). Sensitivity was 95.8% (113/
118), specificity 99.4% (168/169), positive predictive value
Table 1 Results of the first reading at the
test laboratory relating to the final result of
the controller
Results of first reading of
slides at test laboratory
Results of reading of slidesconducted by controller (final
results)
TotalPositive Negative
Positive 86 17 103
Negative 32 152 184
Total 118 169 287
Table 2 Percentage of total false negatives, high false negatives, low false negatives, total false positives and high false positives at the testlaboratory
Category of error Total false negatives High false negatives Low false negatives Total false positives High false positives
Assurance de qualite externe pour le laboratoire de diagnostic de l’ulcere de Buruli au Ghana
objectif Afin d’assurer la qualite du diagnostic de laboratoire pour l’ulcere de Buruli (UB), la microscopie et la PCR ont ete soumises a une assurance
de qualite externe.
methodes Les lames ont ete lues par le personnel de laboratoire et relues de facon aveugle par le verificateur. Parallelement, un test des echantillons
par PCR a ete realise dans le laboratoire de reference local et externe. Les lames et PCR avec des resultats discordants ont ete soumises a une seconde
lecture/test par le verificateur afin de determiner le resultat final. Dans un but de formation du personnel, lames et echantillons PCR ont ete de nouveau
relues/re-testes sous supervision dans le laboratoire controle.
conclusions L’assurance de qualite externe a identifie des deficiences dans la performance du laboratoire. Des actions correctives ont consiste a la
formation sur place ce qui a reduit le nombre de resultats faux negatifs et faux positifs pour la microscopie et la PCR.
mots cles diagnostic de l’ulcere de Buruli, assurance de qualite, microscopie, PCR, Ghana
Control de calidad externo para el diagnostico en laboratorio de la Ulcera de Buruli en Ghana
objetivo Con el fin de asegurar la calidad del diagnostico de laboratorio de la ulcera de Buruli (UB), se sometio a un control de calidad externo (CCE)
la microscopıa y la PCR
metodos Las laminas fueron leıdas por el personal de laboratorio y releıdas por un controlador ciego al resultado previo. Paralelamente se analizaron
las muestras mediante PCR, tanto a nivel local como en un laboratorio de referencia externo. Las laminas y muestras de PCR con resultados
discordantes se sometieron a una segunda lectura/prueba que llevo a cabo el controlador, determinando el resultado final. Las laminas/muestras de PCR
con resultados discrepantes fueron, con fines formativos, releıdas/analizadas por el personal local bajo supervision.
resultados Microscopıa. Primera lectura: tasa de concordancia 82.9%, tasa de discordancia 17.1%, porcentaje de falsos negativos 27.1% (sensi-
bilidad 72.9%), porcentaje de falsos positivos 10.1% (especificidad 89.9%). Segunda lectura: tasa de concordancia 97.9%, tasa de discordancia 2.1%,
porcentaje de falsos negativos 4.2% (sensibilidad 95.8%), porcentaje de falsos positivos 0.6% (especificidad 99.4%). PCR. Primer analisis: tasa de
concordancia 87.9%, tasa de discordancia 12.1%, porcentaje de falsos negativos s 8.2% (sensibilidad 91.8%), porcentaje de falsos positivos 19.1%
(especificidad 80.9%). Segundo analisis: tasa de concordancia 96.2%, tasa de discordancia 3.8%, porcentaje de falsos negativos 4.7% (sensibilidad
95.3%), porcentaje de falsos positivos 2.1% (especificidad 97.9%).
conclusiones El CCE identifico deficiencias en el desempeno del laboratorio. La accion correctiva consistio en entrenar al personal local, reduciendo
en numero de falsos negativos y falsos positivos por microscopıa y PCR.
palabras clave ulcera de Buruli, diagnostico, control de calidad, microscopıa, PCR, Ghana
Tropical Medicine and International Health volume 11 no 11 pp 1688–1693 november 2006
G. Bretzel et al. External quality assurance for the laboratory diagnosis of BUD
ª 2006 Blackwell Publishing Ltd 1693
A stepwise approach to the laboratory diagnosis of Buruli ulcer
disease
G. Bretzel1,2, V. Siegmund1,2, J. Nitschke1,2, K. H. Herbinger1, W. Thompson3, E. Klutse4, K. Crofts5,
W. Massavon6, S. Etuaful7, R. Thompson8, K. Asamoah-Opare8, P. Racz2, F. Vloten2 van, C. Berberich8,
T. Kruppa8, E. Ampadu9, B. Fleischer2 and O. Adjei8
1 Department of Infectious Diseases and Tropical Medicine (DITM), University of Munich, Munich, Germany2 Bernhard Nocht Institute for Tropical Medicine (BNITM), Hamburg, Germany3 Agogo Presbyterian Hospital, Agogo, Ghana4 Dunkwa Governmental Hospital, Dunkwa-on-Offin, Ghana5 Humanitarian Aid Relief Team (HART), Provo, UT, USA6 Goaso Hospital, Goaso, Ghana7 St Martin’s Catholic Hospital, Agroyesum, Ghana8 Kumasi Centre for Collaborative Research in Tropical Medicine (KCCR), Kumasi, Ghana9 National Buruli Ulcer Control Programme (NBUCP), Accra, Ghana
Summary objective In view of technical and financial limitations in areas of endemicity, the current practice and
recommendations for the laboratory diagnosis of Buruli ulcer disease (BUD) may have to be
reconsidered. We reviewed diagnostic results in order to explore options for a modified, more
practicable, cost-effective and timely approach to the laboratory diagnosis of BUD.
methods Diagnostic specimens from 161 clinically diagnosed BUD patients from four different
treatment centres in Ghana were subjected to laboratory analysis. The positivity rates of the laboratory
assays were compared.
results The number of laboratory-confirmed clinically diagnosed BUD cases with one positive con-
firmative test was 20% higher than that with two positive confirmative tests. The specificity of micr-
oscopy (MIC) and PCR was 96.6% and 100%, respectively. Subsequent analysis of specimens from
surgically excised pre-ulcerative tissue-by-tissue MIC and tissue PCR rendered 65% laboratory-con-
firmed BUD cases. Subsequent analysis of diagnostic swabs from ulcerative lesions by swab smear MIC
and swab PCR rendered 70% of laboratory-confirmed BUD cases.
conclusions The specificity of the diagnostic tests used in this study suggests that one positive diag-
nostic test may be considered sufficient for the laboratory confirmation of BUD. Subsequent application
of different diagnostic tests rendered a laboratory confirmation of 65% pre-ulcerative and of 70%
ulcerative lesions. Implementation of a stepwise, subsequent analysis of diagnostic specimens will result
in considerable cost saving compared with simultaneous testing of specimens by several diagnostic assays.
keywords Buruli ulcer, PCR, laboratory diagnosis, reduction of costs, Ghana
Introduction
The laboratory confirmation of Buruli ulcer disease (BUD),
caused by Mycobacterium ulcerans, can be achieved by
analysis of swab and tissue specimens by means of
microscopy (MIC), culture, PCR and histopathology.
According to current WHO recommendations, a positive
diagnosis requires two positive laboratory tests (WHO
2001). In practice, however, these recommendations are
difficult to realize for several reasons.
Previously available data comparing different diagnostic
tests applied on different diagnostic specimens suggest a
wide range of diagnostic sensitivities, varying from as low
as < 40% (swab smear MIC) to > 95% (PCR and
histopathological analysis of tissue specimens) (Guimaraes-
Peres et al. 1999; N’Guessan et al. 2001; WHO 2001;
Whitney et al. 2002). A combination of highly sensitive
tests (PCR and histopathology) performed on invasively
collected tissue specimens would render the two required
positive results with the highest probability. However, the
availability of sophisticated laboratory assays in areas of
endemicity is limited. In general, diagnostic services for
BUD are largely restricted to reference centres. Further-
more, technical difficulties, such as the instability of
Tropical Medicine and International Health doi:10.1111/j.1365-3156.2006.01761.x
volume 12 no 1 pp 89–96 january 2007
ª 2006 Blackwell Publishing Ltd 89
reagents under tropical conditions, hamper molecular
laboratory tests. Histopathology is rarely available in
resource poor countries.
In addition, in most cases, close links between reference
laboratories and peripheral treatment centres do not exist.
Thus, testing of specimens in reference centres may not be
performed on a regular basis. If carried out, the turnaround
time for results may be too long to allow a pre-surgical
diagnosis. Thus, diagnostic information is available only
after the patients have already been treated surgically.
Only a few district level laboratories located in major
treatment centres have started out doing swab smear MIC
for the laboratory diagnosis of BUD. Laboratory capacity
at district level is limited. Diagnostic tests available – if any
– are restricted to MIC. Therefore, on-site testing of
specimens with several assays may not be possible.
In general, restricted health budgets often prevent a
comprehensive laboratory analysis. In view of these limi-
tations, the current diagnostic practice and recommenda-
tions, i.e. simultaneous testing of diagnostic specimens
with different laboratory methods in order to obtain two
positive results, may have to be reconsidered. In order to
strengthen the diagnostic capacity in 2003, a diagnostic
network was initiated in Ghana with financial support of
the Volkswagen foundation. The network initially com-
prised two major treatment centres, one local and two
external reference laboratories providing all diagnostic
laboratory assays and quality assurance. A close cooper-
ation with the National Buruli Ulcer Control Programme
in Ghana has been established. The collaboration has been
extended to other Ghanaian BUD treatment centres. At the
outset of the project, a dry-reagent-based PCR (DRB-PCR)
adapted to tropical conditions was implemented at the
local reference laboratory (Siegmund et al. 2005). Diag-
nostic specimens were subjected to MIC, PCR and culture
(local reference laboratory), as well as histopathology and
Cost calculations for the stepwise approach are based on the assumption that 30% of specimens can be diagnosed by swab smear MIC, an
additional 40% by swab PCR. For the remaining 30%, two options are shown: (a) tissue MIC (additional 5% positive results) followed bytissue PCR (additional 3% positive results) and histopathology (22% remaining), (b) only histopathology. Costs per specimen/test: MIC:
€0.5, PCR: €15, histopathology: €5
Tropical Medicine and International Health volume 12 no 1 pp 89–96 january 2007
G. Bretzel et al. Diagnosing buruli ulcer
94 ª 2006 Blackwell Publishing Ltd
and feasibility of diagnostic methods is required. In this
study a stepwise diagnostic approach, i.e. the subsequent
application of different diagnostic assays, focusing on pre-
surgical diagnosis was evaluated.
The number of laboratory-confirmed clinically diag-
nosed BUD cases with one positive confirmative test was
20% higher than that with two or more positive confirm-
ative tests. In this study, the specificity of PCR was
determined to be 100%. Previous histopathological ana-
lysis of PCR-positive specimens also resulted in a specificity
of 100% (Bretzel et al. 2005). Therefore, single positive
PCR results guarantee a reliable diagnosis of BUD. The
specificity of swab smear and tissue MIC may not be
considered sufficient for the detection of M. ulcerans. In
this study, however, the specificity of MIC as determined
by additional confirmatory tests was 96.6% (57/59). With
1.8% (2/114) false positives related to the total number of
positive patients, the authors consider one diagnostic test
sufficient for the laboratory confirmation of BUD (van der
Werf et al. 2005).
Based on the diagnostic yield of the laboratory assays
analysed in this study, a stepwise application of diagnostic
tests is recommended. Tissue specimens from pre-ulcerative
lesions should first be subjected to tissue MIC (40%
positive results). If MIC is negative, subsequently tissue
PCR (additional 25% positive results) should be carried
out. According to the data presented in this study, the
combination of these methods allows the laboratory
confirmation of approximately 65% of clinically diagnosed
BUD cases with pre-ulcerative lesions. In view of the safety
of laboratory personnel, grinding of tissue for tissue MIC
should be carried out under a laminar flow. Alternatively,
all specimens may be subjected to PCR. Only histopath-
ological analysis can achieve a diagnosis for the remaining
35% of cases.
For patients with ulcerative lesions, diagnostic swabs
should first be examined by swab smear MIC (30%
positive results). If MIC is negative, subsequently swab
PCR (additional 40% positive results) should be carried
out. According to the data presented in this study, the
combination of these methods allows non-invasive pre-
surgical diagnosis of approximately 70% of clinically
diagnosed BUD cases with ulcerative lesions.
Laboratory confirmation of the remaining 30% of cases
may be attempted by post-surgical analysis of excised
tissue by MIC and PCR. However, considering the low
additional diagnostic yield of tissue analysis, PCR analysis
of tissue must be carefully considered in view of the
laboratory costs. Therefore, analysis of the remaining
30% of cases by histopathology may be advisable, if links
with a reference centre providing histopathology are
established.
Implementation of a stepwise, subsequent analysis of
diagnostic specimens will result in considerable cost saving.
Compared with simultaneous testing of diagnostic swabs
and tissue specimens by MIC, PCR and histopathology,
stepwise, successive testing of specimens saves up to 35%
of diagnostic costs in pre-ulcerative lesions and up to 60%
of diagnostic costs in ulcerative lesions.
For pre-surgical diagnosis in pre-ulcerative lesions and in
ulcerative lesions with negative swab samples, punch
biopsies may be used (Phillips et al. 2005). However, the
unit costs of €2 to €3 per biopsy needle must be considered.
For very small pre-ulcerative lesions, punch biopsies may
not be appropriate, as the surface area of the lesion might
not be sufficient to take specimens for several diagnostic
tests.
Timely on-site laboratory diagnosis of BUD requires the
availability of laboratory capacity at district level. In
collaboration with the National Buruli Ulcer Control
Programme in Ghana, decentralization of diagnostic
facilities is underway. BUD MIC and DRB-PCR laborat-
ories have been established in selected treatment centres
at district level, laboratory staff is being trained at the
local reference laboratory, and an external quality assur-
ance system for these laboratories has been installed.
Large-scale on-site assessment of diagnostic specimens
from selected BUD treatment centres will further explore
the practicability and cost-effectiveness of the proposed
stepwise approach to the laboratory diagnosis of BUD.
Acknowledgements
This work was supported by the Volkswagen foundation
(grant number Az.I/79 314). The authors are grateful to
Anja Schorle, Petra Meyer and Birgit Raschdorff for
excellent technical assistance. The authors thank Christel
Schmetz, Erna Fleischmann and Benjamin Plotkin for
critical technical and linguistic review of this article.
References
Bretzel G, Siegmund V, Racz P et al. (2005) Post-surgical assess-
ment of excised tissue from patients with Buruli ulcer disease:
progression of infection in macroscopically healthy tissue.
Tropical Medicine and International Health 10, 1199–1206.
Guarner J, Bartlett J, Whitney EA et al. (2003) Histopathologic
features of Mycobacterium ulcerans infection. Emerging Infec-
tious Diseases 9, 651–656.
Guimaraes-Peres A, Portaels F, de Rijk P et al. (1999) Comparison
of two PCRs for detection of Mycobacterium ulcerans. Journal
of Clinical Microbiology 37, 206–208.
N’Guessan K, Kouassi Y, Bouzid S et al. (2001) [Value and limits
of microscopy of exudates in Mycobacterium ulcerans cutane-
Tropical Medicine and International Health volume 12 no 1 pp 89–96 january 2007
G. Bretzel et al. Diagnosing buruli ulcer
ª 2006 Blackwell Publishing Ltd 95
ous infection in Cote d’Ivoire]. Bulletin de la Societe de
Pathologie Exotique 94, 9–10.
Phillips R, Horsfield C, Kuijper S et al. (2005) Sensitivity of PCR
targeting the IS2404 insertion sequence of Mycobacterium ulc-
erans in an assay using punch biopsy specimens for diagnosis of
Buruli ulcer. Journal of Clinical Microbiology 43, 3650–3656.
Siegmund V, Adjei O, Racz P et al. (2005) Dry-reagent-based PCR
as a novel tool for laboratory confirmation of clinically diag-
nosed Mycobacterium ulcerans-associated disease in areas in the
tropics where M. ulcerans is endemic. Journal of Clinical
Microbiology 43, 271–276.
Stinear T, Ross BC, Davies JK et al. (1999) Identification and
characterization of IS2404 and IS2606: two distinct repeated
sequences for detection of Mycobacterium ulcerans by PCR.
Journal of Clinical Microbiology 37, 1018–1023.
van der Werf TS, Stienstra Y, Johnson RC et al. (2005) Myco-
bacterium ulcerans disease. Bulletin of the World Health
Organization 83, 785–791.
Whitney EAS, Phelan M, Raghunathan PL et al. (2002) Latent
class analysis (LCA) evaluation of four diagnostic tests for
buruli ulcer disease. 5th WHO Advisory Group Meeting on
Buruli Ulcer. WHO, Geneva.
WHO (2001) Buruli Ulcer. Diagnosis of Mycobacterium ulcerans
disease. WHO, Geneva.
Corresponding Author Gisela Bretzel, Department of Infectious Diseases and Tropical Medicine, Ludwig Maximilians University of
Une approche par etape pour le diagnostic de laboratoire de l’ulcere de Buruli
objectifs Vu les limitations techniques et financieres dans les zones endemiques, la pratique actuelle et les recommandations pour le diagnostic de
laboratoire de la maladie de l’ulcere de Buruli (UB) pourraient etre a revoir. Nous avons analyse des resultats de diagnostic afin de d’investiguer des
options pour un diagnostic de laboratoire modifie pour l’UB qui soit plus pratique, avec un bon rapport cout/efficacite et une approche plus opportune.
methodes Des specimens de 161 patients UB cliniquement diagnostiques dans quatre centres de traitement differents du Ghana ont ete analyses en
utilisant des tests de laboratoire et les taux de positivite ont ete compares.
resultats Le nombre de cas d’UB cliniquement diagnostiques et confirmes par un seul test positif de laboratoire etait 20% plus eleve que celui par
deux tests positifs de laboratoire. La specificite de la microscopie et de la PCR etait de 96,6% et 100% respectivement. Une analyse subsequente par
microscopie et par PCR, de specimens provenant d’excision de tissu en state pre ulcere a revele 65% de cas d’UB confirmes au laboratoire. L’analyse
subsequente par microscopie et par PCR sur des prelevements sur tiges de coton de lesions ulcerees a revele 70% de cas d’UB confirmes au laboratoire.
conclusions La specificite des tests de diagnostic utilises dans cette etude suggere qu’un seul test positif peut etre considere suffisant pour la
confirmation de laboratoire de l’UB. L’application subsequente de differents tests de laboratoire a permis une confirmation de laboratoire pour 65% de
cas de lesions en stade pre ulcere et 70% de cas de lesions ulcerees. L’implementation d’une analyse par etape des specimens permettra de reduire
considerement les couts par rapport au test simultane des specimens par plusieurs tests de diagnostic.
mots cles ulcere de Buruli, PCR, microscopie, reduction des couts, Ghana
Enfoque escalonado del diagnostico en laboratorio de la ulcera de Buruli
objetivo En vista de las limitaciones tecnicas y economicas existentes en las areas endemicas, deberıa reconsiderarse la practica y recomendaciones
actuales para el diagnostico en laboratorio de la ulcera de Buruli (UB). Hemos revisado los resultados del diagnostico en laboratorio de la UB, con el fin
de explorar opciones para un nuevo enfoque mas practico, costo-efectivo y oportuno.
metodos Se analizaron 161 especimenes de pacientes con diagnostico clınico de UB de cuatro centros diferentes en Ghana. Se comparo la tasa de
positividad de las muestras.
resultados El numero de casos clınicos de UB confirmados por laboratorio con una prueba positiva confirmatoria fue un 20% mas alta que con dos
pruebas confirmatorias. La especificidad de la microscopıa y la PCR fue del 96.6% y 100% respectivamente. Analisis subsiguientes de especimenes
diagnosticos de frotis de lesiones ulcerosas mediante microscopıa y PCR dieron un 70% de los casos de UB previamente confirmados en el laboratorio.
conclusiones La especificidad de las pruebas diagnosticas utilizadas en este estudio sugiere que una prueba diagnostica positiva puede considerarse
como suficiente para la confirmacion en laboratorio de la UB. La aplicacion subsiguiente de diferentes tecnicas diagnosticas tuvo una confirmacion del
65% para lesiones pre-ulcerativas y un 70% para las ulcerativas. La implementacion de un analisis escalonado de los especimenes diagnosticos resultarıa
en un ahorro considerable, comparado con una evaluacion simultanea de los especimenes utilizando diferentes pruebas diagnosticas.
palabras clave Ulcera de Buruli, PCR, microscopıa, ahorro, Ghana
Tropical Medicine and International Health volume 12 no 1 pp 89–96 january 2007
G. Bretzel et al. Diagnosing buruli ulcer
96 ª 2006 Blackwell Publishing Ltd
68 • CID 2007:45 (1 July) • Siegmund et al.
M A J O R A R T I C L E
Dry Reagent–Based Polymerase Chain ReactionCompared with Other Laboratory Methods Availablefor the Diagnosis of Buruli Ulcer Disease
Vera Siegmund,1,2 Ohene Adjei,3 Jorg Nitschke,2 William Thompson,4 Erasmus Klutse,5 Karl Heinz Herbinger,1
Ruth Thompson,3 Felicitas van Vloten,2 Paul Racz,2 Bernhard Fleischer,2 Thomas Loescher,1 and Gisela Bretzel1,2
1Department of Infectious Diseases and Tropical Medicine, Ludwig-Maximilians University of Munich, Munich, and 2Bernhard Nocht Institute forTropical Medicine, Hamburg, Germany; and 3Kumasi Centre for Collaborative Research in Tropical Medicine, Kwame Nkrumah University ofScience and Technology, Kumasi, 4Agogo Presbyterian Hospital, Agogo, and 5Dunkwa Governmental Hospital, Dunkwa-on-Offin, Ghana
Background. Because of the multifaceted clinical presentation of Buruli ulcer disease, misclassification ofclinically diagnosed cases may occur frequently. Laboratory tests for the confirmation of suspected cases includemicroscopic examination, culture, polymerase chain reaction (PCR), and histopathologic examination. However,microscopic examination, the only test usually available in areas of endemicity, has a low sensitivity.
Methods. To make a highly sensitive diagnostic method locally available, dry reagent–based PCR (DRB-PCR),which is well adapted to tropical conditions, was pilot-tested in Ghana. Subsequently, the assay was used for theroutine diagnosis of Buruli ulcer disease over a period of 2 years. The method was compared with other diagnostictests to evaluate its performance under field conditions.
Results. The interassay agreement rate between DRB-PCR and standard PCR was 91.7% for swab specimensand 95% for tissue specimens. Among all of the locally available tests, DRB-PCR revealed the highest overallpositivity ratio. Sixty percent of patients with clinical diagnoses of Buruli ulcer disease had the diagnoses confirmedby DRB-PCR of swab or tissue specimens, compared with 30%–40% of patients who had diagnoses confirmedby microscopic examination of swab or tissue specimens. The positivity ratio of DRB-PCR varied considerablywhen analyzed per treatment center. Standardization of specimen collection resulted in a 30% increase in thepositivity ratio of the assay, compared with that in the pilot-testing phase.
Conclusions. DRB-PCR is a reliable tool for the diagnosis of Buruli ulcer disease. However, PCR assays aresuitable for detection only during early stages of the disease, when samples still contain bacilli. The quality ofclinical diagnosis and the quality of diagnostic specimens strongly influence the positivity ratio.
Buruli ulcer disease (BUD) is an infectious disease
caused by Mycobacterium ulcerans that involves the skin
and the subcutaneous adipose tissue. In the majority
of cases, the disease starts as a painless papule, plaque,
or nodule that evolves into a large ulceration with char-
acteristic undermined edges. Destructive lesions with
extensive scarring and contractures of the limbs are
common. Because of the multifaceted clinical presen-
tation, misclassification of clinically diagnosed cases of
Received 28 November 2006; accepted 24 March 2007; electronically published21 May 2007.
Reprints or correspondence: Dr. Vera Siegmund, Dept. of Infectious Diseasesand Tropical Medicine, Ludwig-Maximilians University of Munich, Leopoldstrasse5, D-80802 Munich, Germany ([email protected]).
Clinical Infectious Diseases 2007; 45:68–75� 2007 by the Infectious Diseases Society of America. All rights reserved.1058-4838/2007/4501-0013$15.00DOI: 10.1086/518604
BUD may occur frequently. After tuberculosis and lep-
rosy, BUD has become the third most common my-
cobacterial disease in immunocompetent humans
worldwide, and the incidence of BUD is increasing in
tropical countries. The incidence and prevalence of
BUD worldwide are not precisely defined, because ad-
equate surveillance data based on accurate case confir-
mation data are lacking. Thus far, disease control in
countries where BUD is endemic is limited to early case
detection through improved active surveillance and
surgical excision of lesions followed by skin grafting
[1–6].
Laboratory confirmation of suspected cases prevents
misclassification, ensures adequate treatment, and pro-
vides reliable incidence and prevalence data that, in
turn, may support the development of new disease-
control strategies. Currently available laboratory tests
include microscopic detection of acid-fast bacilli in
DRB-PCR vs. Other Methods for M. ulcerans • CID 2007:45 (1 July) • 69
Table 1. Diagnostic specimens, transportation media, and diagnostic tests.
Specimen, transportation medium Diagnostic test
TissuePANTA, 5 mL Microscopic examination and cultureCLS, 700 mL DRB-PCRCLS, 700 mL Standard PCRTen percent buffered neutral formalin, 5 mL Histopathologic examination
SwabPANTA, 5 mL Microscopic examination and cultureCLS, 700 mL DRB-PCRCLS, 700 mL Standard PCR
Microscopic examinationA 25/61 41.0 36/61 59.0 26/49 53.1 62/110 56.4AG 2/6 33.3 2/6 33.3 2/6 33.3 4/12 33.3D 10/36 27.8 9/36 25.0 10/27 37.0 19/63 30.2F 1/12 8.3 NA NA NA NA NA NAG 3/17 17.6 7/17 41.2 1/16 6.3 8/33 24.2All centers 41/132 31.1 (23.2–39.0) 54/120 45.0 (36.1–53.9) 39/98 39.8 (30.1–49.5) 93/218 42.7 (36.1–49.2)
CultureA 0/61 0 2/61 3.3 4/49 8.2 6/110 5.5AG 0/6 0 1/6 16.7 0/6 0 1/12 8.3D 0/36 0 1/36 2.8 1/27 3.7 2/63 3.2F 0/12 0 NA NA NA NA NA NAG 0/17 0 0/17 0 0/16 0 0/33 0All centers 0/132 0 4/120 3.3 (0.1–6.5) 5/98 5.1 (0.7–9.5) 9/218 4.1 (1.5–6.8)
DRB-PCRA 41/61 67.2 42/61 68.9 37/49 75.5 79/110 71.8AG 4/6 66.7 5/6 83.3 5/6 83.3 10/12 83.3D 23/36 63.9 13/36 36.1 13/27 48.1 26/63 41.3F 4/12 33.3 NA NA NA NA NA NAG 8/17 47.1 10/17 58.8 7/16 43.8 17/33 51.5All centers 80/132 60.6 (52.3–68.9) 70/120 58.3 (49.5–67.2) 62/98 63.3 (53.7–72.8) 132/218 60.6 (54.1–67.0)
Standard PCRA 45/61 73.8 43/61 70.5 35/49 71.4 78/110 70.9AG 4/6 66.7 5/6 83.3 4/6 66.7 9/12 75.0D 24/36 66.7 17/36 47.2 13/27 48.1 30/63 47.6F 4/12 33.3 NA NA NA NA NA NAG 12/17 70.6 10/17 58.8 7/16 43.8 17/33 51.5All centers 89/132 67.4 (59.4–75.4) 75/120 62.5 (53.8–71.2) 59/98 60.2 (50.5–69.9) 134/218 61.5 (55.0–67.9)
Histopathologic examinationA NA NA 38/46 82.6 34/39 87.2 72/85 84.7AG NA NA 5/6 83.3 2/2 100.0 7/8 87.5D NA NA 24/30 80.0 15/24 62.5 39/54 72.2F NA NA NA NA NA NA NA NAG NA NA 7/9 77.8 5/11 45.5 12/20 60.0All centers NA NA 74/91 81.3 (73.3–89.3) 56/76 73.7 (63.8–83.6) 130/167 77.8 (71.5–84.1)
NOTE. A, Agogo; AG, Agroyesum; D, Dunkwa; DRB-PCR, dry reagent–based PCR; F, Field; G, Goaso; NA, not available.
DRB-PCR vs. Other Methods for M. ulcerans • CID 2007:45 (1 July) • 73
Table 5. Results of dry reagent–based PCR (DRB-PCR) versushistopathologic examination.
DRB-PCR result
Histopathologic examination result
Total
BUD(n p 130)
Not BUD(n p 37)
AFBpositive
AFBnegative
AFBnegative
Positive 89 14 1 104Negative 16 11 36 63
Total 105 25 37 167
NOTE. Data are no. of specimens. AFB, smear for acid-fast bacilli; BUD,Buruli ulcer disease.
abscess; 2 patients had chronic or bacterial ulceration of the
skin; 8 patients had parasitic infections [5 of these infections
were onchocerciasis]; 2 patients had nonspecific inflammation;
9 patients had nonspecific, chronic, or acute dermatitis; 1 pa-
tient had fibrosis; and 1 patient had an enlarged lymph node).
DRB-PCR versus histopathologic examination. Table 5
compares the results of DRB-PCR with those of histopathologic
examination. Of the 130 specimens confirmed by histopatho-
logic examination, 105 were smear positive for acid-fast bacilli.
DRB-PCR yielded positive results for 89 of these specimens
and negative results for 16 of these specimens. Of the 37 spec-
imens that were negative by histopathologic examination, 1
specimen was positive by DRB-PCR.
DISCUSSION
To facilitate the application of PCR for the diagnosis of BUD
in countries where the disease is endemic, a DRB-PCR was
installed at the Kumasi Centre for Collaborative Research in
Tropical Medicine. The method was used under routine con-
ditions for 2 years and compared with other diagnostic tests,
including standard PCR.
Although the reagents are slightly more expensive, DRB-PCR
offers considerable advantages over conventional PCR assays
for application in resource-poor countries. The use of lyo-
philized reagents, which are stable at room temperature for at
least 1 year, facilitates shipment and storage. The assay is less
sensitive to tropical climatic conditions, such as heat and hu-
midity, as well as the freezing and thawing effects that are
consequences of frequent power outages. The reduced number
of work steps shortens the performance time and minimizes
the risk of contamination. The simplified handling of the test
allows rapid familiarization with the method; laboratory staff
without previous PCR training were capable of performing the
assay correctly after !1 week.
The overall agreement rates between DRB-PCR and standard
PCR of 91.7% for swab specimens and 95.0% for tissue spec-
imens were similar to those obtained in the pilot study [7].
The observed deviations of up to 8% between the 2 assays are
probably because of the use of different swab and tissue spec-
imens for PCR analysis at the Kumasi Centre for Collaborative
Research and at the Department of Infectious Diseases and
Tropical Medicine at the Ludwig-Maximilians University of
Munich. Although adjacent tissue specimens were subjected to
both methods, the possibility of one specimen containing fewer
bacilli than the other specimen or containing no bacilli cannot
be excluded. Likewise, if several swab specimens are collected
for different tests, the second or third swab specimen is likely
to contain fewer bacilli than the first specimen. In addition, it
is well known that, even if identical source material is used,
repeated testing of material containing few bacilli may render
variable results. In view of the agreement rates, DRB-PCR is a
reliable method for the laboratory diagnosis of BUD [7, 13].
When analyzed according to the origin of specimens, the
agreement rates differed among hospitals, with the highest in-
terassay agreement rates being detected at centers A and D
(table 3). The specimen collection guidelines were developed
in close collaboration with these treatment centers, training of
hospital staff was performed at least twice yearly by staff from
the Department of Infectious Diseases and Tropical Medicine
at the Ludwig-Maximilians University of Munich, and regular
technical assistance for specimen collection was provided by
staff from the Kumasi Centre for Collaborative Research. Be-
cause of the remoteness of the other centers and the irregular
patient intake, specimens were submitted without regular as-
sistance from laboratory staff. Therefore, we conclude that the
quality of laboratory diagnosis strongly depends on the quality
of specimens and the training status of the hospital staff col-
lecting the specimens.
The overall positivity ratio of DRB-PCR of swab specimens
(60.6%) was nonsignificantly lower than the overall positivity
ratio of standard PCR of swab specimens (67.4%). Overall
positivity ratios for tissue specimens (60.6% by DRB-PCR and
61.5% by standard PCR) were comparable. Among all tests
locally available at the Kumasi Centre for Collaborative Re-
ratios (60.6% for swab and tissue specimens) than did micro-
scopic examination (31.1% for swab specimens and 42.7% for
tissue specimens) and culture (0% for swab specimens and
4.1% for tissue specimens). In general, compared with the pilot
study, the overall positivity ratio of DRB-PCR of swab and
tissue specimens increased by 30% [7]. This increase can be
mainly attributed to the patient inclusion criteria in this study
(duration of lesions of !6 months, compared with all stages of
the disease in the pilot-study patients) and the standardized
specimen collection.
The low overall positivity ratio for cultures was likely to be
caused by presurgical, short-term antimycobacterial treatment,
which was common practice in Ghana during the study period.
Because of the long generation time of M. ulcerans, culture, in
74 • CID 2007:45 (1 July) • Siegmund et al.
general, cannot be considered to be a tool for the timely di-
agnosis of BUD.
The positivity ratio of each test method showed considerable
variation when analyzed per hospital. Also, in that respect, the
positivity ratio of a test depends on the quality of the diagnostic
specimen submitted for analysis. In general, the collection of
diagnostic swab specimens (circling the entire undermined edge
of the lesion) is less error-prone than the collection of diag-
nostic tissue specimens. The tissue specimen has to be obtained
from the edge of the lesions below the end of the undermined
edge containing necrotic tissue sections and the subcutaneous
adipose tissue, where the bacilli are present. In specimens col-
lected from excised tissue, the correct location is difficult to
define, because the end of the undermined edge may not be
clearly recognizable after excision. Therefore, errors may occur.
positivity ratios of DRB-PCR of swab specimens from hospitals
A (67.2%) and D (63.9%) were comparable. However the pos-
itivity ratio of the corresponding ulcerative tissue specimens
from hospital D (36.1%) was considerably lower than that for
hospital A (68.9%). These findings suggest that the collection
of swab specimens did not pose any difficulties in either of
these hospitals; however, problems with the collection of tissue
specimens may have occurred in hospital D.
In addition, the quality of the clinical diagnosis, and, thus,
the selection of patients for laboratory diagnosis, strongly in-
fluence the positivity ratio of a test. Onchocerciasis nodules,
for example, may be frequently misclassified as BUD. Table 4
shows that only 48% of the preulcerative specimens from hos-
pital D, which is located in an area where onchocerciasis is
endemic, were confirmed by PCR. Histopathologic analysis of
the preulcerative specimens that had negative PCR results re-
vealed that 18.5% of the preulcerative lesions from hospital D
were, in fact, onchocerciasis nodules. Onchocerciasis did not
occur in the preulcerative specimens from hospital A, which is
not located in an area of endemicity, and 75% of these spec-
imens were confirmed by PCR.
Histopathologic examination confirmed 77.8% of all sus-
pected cases of BUD. The majority of lesions from patients
with positive histopathologic results and negative PCR results
progressed toward the healing stage, thus containing few or no
bacilli. This finding reflects the major limitation of diagnostic
PCRs leading to false-negative results, compared with histo-
pathologic examination. PCR is a suitable diagnostic tool only
for early stages of the disease, when sufficient amounts of bacilli
are present. However, false-negative histopathologic examina-
tion results also occur. One patient who presented with the
typical clinical picture of BUD, positive results of microscopic
examination of swab specimens, and positive results of PCR of
swab and tissue specimens was not considered to have BUD
by histopathologic examination. Thus, a true gold standard for
the laboratory confirmation of BUD does not exist.
The specificity of the IS2404 PCR for clinically relevant my-
cobacteria was investigated elsewhere [7, 12]. The only my-
cobacterial species detectable with IS2404 PCR, Mycobacterium
liflandii, is present only in environmental samples [16]. There-
fore, false-positive PCR reactions in clinical samples do not
occur.
The results of this study show that DRB-PCR is a reliable
tool for the laboratory confirmation of cases clinically diag-
nosed as BUD by analysis of swab and tissue specimens. How-
ever, the performance of the assay is clearly linked to the quality
of clinical diagnosis and specimen collection. To grant reliable
results, the laboratory should not limit its activities to the lab-
oratory work but should collaborate closely with the treatment
centers with regard to training and supervision of specimen
collection.
Acknowledgments
We thank Anja Schorle, Petra Meyer, and Birgit Raschdorff (BernhardNocht Institute for Tropical Medicine, Hamburg, Germany); Erna Fleisch-mann (Department of Infectious Diseases and Tropical Medicine, Ludwig-Maximilians University of Munich, Munich, Germany); and Leticia Kunaa(Kumasi Centre for Collaborative Research, Kumasi, Ghana), for excellenttechnical assistance.
Potential conflicts of interest. All authors: no conflicts.
References
1. Sizaire V, Nackers F, Comte E, Portaels F. Mycobacterium ulcerans in-fection: control, diagnosis, and treatment. Lancet Infect Dis 2006; 6:288–96.
2. van der Werf TS, Stienstra Y, Johnson RC, et al. Mycobacterium ulceransdisease. Bull World Health Organ 2005; 83:785–91.
3. World Health Organization. Buruli ulcer. Management of Mycobac-terium ulcerans disease. Geneva: World Health Organization, 2001.
4. Amofah G, Bonsu F, Tetteh C, et al. Buruli ulcer in Ghana: results ofa national case search. Emerg Infect Dis 2002; 8:167–70.
5. Debacker M, Aguiar J, Steunou C, et al. Mycobacterium ulcerans disease(Buruli ulcer) in rural hospital, southern Benin, 1997–2001. EmergInfect Dis 2004; 10:1391–8.
6. King CH, Ashford DA, Dobos K, et al. Mycobacterium ulcerans infectionand Buruli ulcer disease: emergence of a public health dilemma. Wash-ington, DC: ASM Press, 2001.
7. Siegmund V, Adjei O, Racz P, et al. Dry-reagent-based PCR as a noveltool for laboratory confirmation of clinically diagnosed Mycobacteriumulcerans–associated disease in areas in the tropics where M. ulcerans isendemic. J Clin Microbiol 2005; 43:271–6.
8. Whitney EAS, Phelan M, Raghunathan PL, et al. Latent class analysis(LCA) evaluation of four diagnostic tests for Buruli ulcer disease. In:5th World Health Organization Advisory Group Meeting on BuruliUlcer. Geneva: World Health Organization, 2002.
9. World Health Organization. Buruli ulcer. Diagnosis of Mycobacteriumulcerans disease. Geneva: World Health Organization, 2001.
10. Guimaraes-Peres A, Portaels F, de Rijk P, et al. Comparison of twoPCRs for detection of Mycobacterium ulcerans. J Clin Microbiol1999; 37:206–8.
11. N’Guessan K, Kouassi Y, Bouzid S, et al. Value and limits of microscopyof exudates in Mycobacterium ulcerans cutaneous infection in Coted’Ivoire [in French]. Bull Soc Pathol Exot 2001; 94:9–10.
12. Stinear T, Ross BC, Davies JK, et al. Identification and characterization
DRB-PCR vs. Other Methods for M. ulcerans • CID 2007:45 (1 July) • 75
of IS2404 and IS2606: two distinct repeated sequences for detectionof Mycobacterium ulcerans by PCR. J Clin Microbiol 1999; 37:1018–23.
13. Bretzel G, Siegmund V, Nitschke J, et al. External quality assurance forthe laboratory diagnosis of Buruli ulcer disease in Ghana. Trop MedInt Health 2006; 11:1688–93.
14. Bretzel G, Siegmund V, Nitschke J, et al. A stepwise approach to thelaboratory diagnosis of Buruli ulcer disease. Trop Med Int Health2007; 12:89–96.
15. Guarner J, Bartlett J, Whitney EA, et al. Histopathologic features ofMycobacterium ulcerans infection. Emerg Infect Dis 2003; 9:651–6.
16. Mve-Obiang A, Lee RE, Umstot ES, et al. A newly discovered myco-bacterial pathogen isolated from laboratory colonies of Xenopus specieswith lethal infections produces a novel form of mycolactone, the My-cobacterium ulcerans macrolide toxin. Infect Immun 2005; 73:3307–12.
Different Diagnostic Methods for BUD • CID 2009:48 (15 April) • 1055
M A J O R A R T I C L E
Comparative Study of the Sensitivity of DifferentDiagnostic Methods for the Laboratory Diagnosisof Buruli Ulcer Disease
Vera Siegmund,1 Jorg Nitschke,1,3 William Thompson,6 Erasmus Klutse,7 Pius Agbenorku,5 Alexander Schipf,2,9
Simone Reu,2 Paul Racz,3 Bernhard Fleischer,3 Marcus Beissner,1 Erna Fleischmann,1 Kerstin Helfrich,1
Tjip S. van der Werf,8 Thomas Loscher,1 and Gisela Bretzel1
Departments of 1Infectious Diseases and Tropical Medicine and 2Pathology, Ludwig-Maximilians University of Munich, Munich, and 3BernhardNocht Institute for Tropical Medicine, Hamburg, Germany; 4Kumasi Centre for Collaborative Research in Tropical Medicine and 5ReconstructivePlastic Surgery and Burns Unit, Department of Surgery, School of Medical Sciences, Kwame Nkrumah University of Science and Technology,Kumasi, 6Agogo Presbyterian Hospital, Agogo, and 7Dunkwa Governmental Hospital, Dunkwa-on-Offin, Ghana; 8University Medical CentreGroningen, University of Groningen, Groningen, The Netherlands; and 9Institute of Pathology, Kantonsspital Luzern, Luzern, Switzerland
Background. Several diagnostic laboratory methods are available for case confirmation of Buruli ulcer disease.This study assessed the sensitivity of various diagnostic tests in relation to clinical presentation of the disease, typeof diagnostic specimen, and treatment history.
Methods. Swab samples, 3-mm punch biopsy tissue specimens, and surgically excised tissue specimens from384 individuals with suspected Buruli ulcer disease were obtained at 9 different study sites in Ghana and wereevaluated with dry reagent–based polymerase chain reaction (PCR), microscopic examination, culture, and his-topathological analysis. The study subjects presented with nonulcerative and ulcerative lesions and were dividedinto 3 treatment groups: (1) previously untreated patients scheduled for antimycobacterial treatment, (2) patientstreated with surgery alone, and (3) patients treated with surgery in combination with previous antimycobacterialtreatment.
Results. Of 384 suspected cases of Buruli ulcer disease, 268 were confirmed by at least 1 positive test result.The overall sensitivity of PCR (85%) was significantly higher than that of microscopic examination (57%) andculture (51%). After data were stratified by treatment group, type of lesion, and diagnostic specimen type, analysisrevealed that PCR of 3-mm punch biopsy tissue specimens (obtained from previously untreated nonulcerativelesions) and of swab samples (obtained from previously untreated ulcers) had the highest diagnostic sensitivity(94% and 90%, respectively). Although duration of the disease did not significantly influence the sensitivity ofany test, previous antimycobacterial treatment was significantly associated with decreased sensitivity of PCR andculture.
Conclusions. Across all subgroups, PCR had the highest sensitivity. PCR assessment of 3-mm punch biopsytissue specimens proved to be the best diagnostic tool for nonulcerative lesions, and PCR assessment of swabsamples was the best diagnostic tool for ulcerative lesions. For monitoring of antimycobacterial treatment successwithin controlled trials, however, only culture is appropriate.
Buruli ulcer disease (BUD), which is caused by My-
cobacterium ulcerans, affects the skin and subcutaneous
adipose tissue. BUD occurs in 130 countries worldwide,
Received 17 August 2008; accepted 14 December 2008; electronically published10 March 2009.
Reprints or correspondence: Dr. Karl-Heinz Herbinger, Dept. of InfectiousDiseases and Tropical Medicine, Ludwig-Maximilians University of Munich,Leopoldstrasse 5, D-80802 Munich, Germany ([email protected]).
Clinical Infectious Diseases 2009; 48:1055–1064� 2009 by the Infectious Diseases Society of America. All rights reserved.1058-4838/2009/4808-0005$15.00DOI: 10.1086/597398
with a focus and an increasing number of cases oc-
curring in West Africa [1–3]. The disease initially pre-
sents as a painless nodule, papule, plaque, or edema
and evolves into a painless ulcer with characteristically
undermined edges. If untreated, scarring and contrac-
tures may cause serious functional disabilities [3, 4].
Previously, BUD was treated with wide surgical exci-
sion; in 2004, however, antimycobacterial treatment
alone or in combination with surgery was introduced
[3–8]. Currently available diagnostic laboratory tests
include microscopic examination, culture, IS2404 PCR
1056 • CID 2009:48 (15 April) • Herbinger et al.
of swab samples and tissue specimens, and histopathological
analysis.
According to various studies of the laboratory confirmation
of clinically suspected cases of BUD, microscopic examination
detects 29%–78% of BUD cases, and culture detects 34%–79%.
Antimycobacterial treatment before specimen collection was
shown to reduce culture positivity ratios (i.e., the number of
positive test results divided by the number of cases of clinically
suspected BUD) to !5%. Histopathological analysis confirmed
190% of clinically diagnosed BUD cases and 170% of clinically
suspected BUD cases [8–15].
With an analytical specificity of 100% [16–19], a sensitivity
of 79%–85% [10,12], and positivity ratios of 61%–72% [14,
15], IS2404 PCR is considered to be the most reliable technique
for the detection of M. ulcerans in human diagnostic samples.
Laboratory assessment of nonulcerative lesions is restricted to
an analysis of tissue specimens, whereas diagnostic swab sam-
ples provide a valuable alternative for assessment of ulcerative
lesions. According to recent data from Ghana, microscopic ex-
amination and/or PCR of swab samples allowed confirmation
of up to 70% of suspected cases of BUD [14, 20].
In 2001, the World Health Organization recommended that
2 positive laboratory test results be obtained to confirm a pos-
itive diagnosis [21]. However, laboratory confirmation of sus-
pected cases of BUD by 1 positive test result yields ∼20% more
confirmed cases than does confirmation by 2 positive test re-
sults. Because of the high positive predictive values of IS2404
PCR (100%) and microscopic examination (97%), 1 positive
test result is considered to be sufficient for confirmation of a
diagnosis of BUD [14, 20]. A positive IS2404 PCR result is also
regarded as adequate evidence to commence antimycobacterial
treatment [22].
In addition to swab samples, punch biopsy tissue specimens
also allow the pretreatment laboratory confirmation of sus-
pected BUD [3, 22]. Data on the diagnostic use of punch biopsy
tissue specimens, however, are still scarce. Phillips et al. [23]
reported sensitivities of 42% for microscopic examination, 49%
for culture, 98% for IS2404 PCR, and 82% for histopathological
analysis of 4-mm and 6-mm punch biopsy tissue specimens.
In the context of a research program funded by the European
Commission on diagnosis and antimycobacterial treatment of
BUD, various types of diagnostic specimens were obtained from
patients who presented with different clinical forms of the dis-
ease and were grouped into 3 different treatment categories.
The aim of this study was to determine the sensitivities of
diagnostic laboratory methods for various types of specimens,
depending on the type of lesions and prior treatment history.
PATIENTS, MATERIALS, AND METHODS
Study population and inclusion criteria. The study included
individuals with clinical suspicion of BUD who had nonulcer-
ative or ulcerative lesions and were seen from January 2006
through February 2008 at 9 different study sites in Ghana. The
study subjects belonged to 1 of 3 different treatment groups:
(1) the drug treatment group, which included patients who
were scheduled for drug treatment, had received no antimy-
cobacterial treatment before specimen collection, had a �6-
month duration of disease, had lesions �10 cm in diameter,
and were �5 years of age; (2) the surgical treatment group,
which included patients who were treated with surgical excision
and had received no previous antimycobacterial treatment; and
(3) the surgical treatment plus antimycobacterial treatment
group, which included patients who were treated with surgical
excision and had received at least 7 days of previous antimy-
cobacterial treatment.
Standardized specimen collection. In the majority of cases,
the diagnostic specimens were collected during the patients’
initial presentation to the hospital. For a limited number of
patients, additional follow-up samples were analyzed. Swab
samples were taken by circling the entire undermined edges of
and azlocillin), and histological specimens were collected in 5
mL of 10% buffered neutral formalin [14, 20].
Diagnostic methods and laboratories. Diagnostic speci-
mens were processed at the Kumasi Centre for Collaborative
Research in Tropical Medicine in Kumasi, Ghana, by dry re-
agent–based IS2404 PCR, microscopic examination, and cul-
ture with use of standardized procedures [14, 19–21]. Standard
IS2404 PCR was performed and slides were reread for external
quality assurance by the Department for Infectious Diseases
and Tropical Medicine at the University of Munich (Munich,
Figure 1. Flow chart of the study procedure. Specimens (swab samples [swab], punch biopsy tissue specimens [tissue-punch], and surgically excisedtissue specimens [tissue-surgery]) obtained from 384 patients with suspected Buruli ulcer disease (BUD) from 3 different treatment groups (drugtreatment [DT], surgical treatment without previous antimycobacterial treatment [ST], and surgical treatment with previous antimycobacterial treatment[ST+]) for BUD diagnostic testing with dry reagent–based IS2404 PCR (PCR), microscopy (MIC), culture with confirmatory IS2404 dry reagent–basedPCR (CUP), and histopathologcial analysis (HIS). There were a total of 268 laboratory-confirmed cases of BUD (A) and 116 suspected cases of BUDwithout laboratory confirmation (B). The statistics are based on the data obtained from each patient’s first lesion (if there was 11 lesion per patient)and each patient’s first visit at hospital (if there was 11 visit per patient). *Four of 16 patients with suspected BUD who had negative findings atinitial presentation (3 in the ST group and 1 in the DT group) had BUD confirmed at follow-up visits. **Twenty-six of 44 patients in the DT groupwith suspected BUD that was not laboratory confirmed (15 patients with nonulcerative lesions and 29 patients with ulcerative lesions) receivedantimycobacterial treatment, because the clinical findings were suggestive of BUD. Three additional patients with suspected BUD in the DT groupreceived treatment for onchocerciasis. The remaining 15 patients in the DT group with suspected BUD were lost to follow-up.
1058 • CID 2009:48 (15 April) • Herbinger et al.
Germany), and histopathological examination was performed
at the Bernhard Nocht Institute for Tropical Medicine (Ham-
burg, Germany) and the Department of Pathology at the Uni-
versity of Munich [11, 17, 24].
In brief, DNA was prepared using the Puregene DNA iso-
lation kit, with minor modifications [19]. For dry reagent–
based PCR, the oligonucleotides MU5 and MU6 [17] were
lyophilized in reaction tubes. PuReTaqTM Ready-To-GoTM
PCR Beads (Amersham Biosciences) were added and dissolved
in water before adding the template DNA. The standard PCR
was performed according to the protocol described by Stinear
et al. [17]. Both PCR assays included negative extraction and
positive, negative, and inhibition controls.
Culture specimens were decontaminated by the Petroff
method, inoculated on Loewenstein-Jensen media, and incu-
bated at 32�C for 6 months. Microscopy smears were prepared
from decontaminated material and were stained with the Ziehl-
Neelsen technique [22].
Cultures with growth were subjected to Ziehl-Neelsen stain-
ing and a confirmatory IS2404 PCR. If a negative PCR result
was obtained, sequence analysis of the rpoB gene (342 base
was used for patients who received drug treatment and for
patients who underwent surgical procedures.
RESULTS
Subjects. Four hundred sets of specimens from 384 individ-
uals with suspected BUD (drug treatment group, 204 patients;
surgical treatment group, 109 patients; and surgical treatment
plus antimycobacterial treatment group, 71 patients) who pre-
sented with 140 nonulcerative (35%) and 260 ulcerative (65%)
lesions were collected at the first presentation at the hospital.
Sixteen patients had 2 lesions. A total of 43 sets of follow-up
specimens were obtained from 30 study subjects; however, only
data from the patients’ initial specimen collection were con-
sidered for analysis. Forty-eight percent of the suspects were
5–14 years of age (age range, 1–95 years; mean age, 21 years;
median age, 14 years); 40.9% of the patients were male (figure
1).
Laboratory-confirmed BUD cases. One hundred seventy-
two (44%) of the BUD cases were confirmed by at least 2
positive laboratory test results; 268 (69.8%) were confirmed by
at least 1 positive laboratory test result (figure 1). One hundred
fourteen (42.5%) of the patients with BUD presented with
nonulcerative lesions, and 154 (57.5%) presented with ulcer-
ative lesions; 108 (40.3%) of the patients were male, and 150
(56.6%) of 265 patients (for 3 patients, age was unknown) were
5–14 years of age (range, 2–80 years; mean age, 18 years; median
age, 12 years) (figure 2). In 231 (93.1%) of 248 patients, the
lesions were located on the limbs or shoulders, with the right
side being affected statistically significantly more often (in 138
[59.7%] of 231 patients; ) than the left side (93P p .035
[40.3%] of 231 patients) (table 1). The lesions of 243 patients
with confirmed cases with known lesion sizes were distributed
according to World Health Organization categories [3], as fol-
lows: category I (a single lesion !5 cm in diameter), 108 patients
(44.4%), including 61 in the drug treatment group, 23 in the
surgical treatment group, and 24 in the surgical treatment plus
antimycobacterial treatment group; category II (a single lesion
5–15 cm in diameter), 127 patients (52.3%), including 93 in
the drug treatment group, 15 in the surgical treatment group,
and 19 in the surgical treatment plus antimycobacterial treat-
ment group; and category III (a single lesion 115 cm in di-
ameter, multiple lesions, or osteomyelitis), 8 patients (3.3%),
including 4 in the drug treatment group, 3 in the surgical
treatment group, and 1 in the surgical treatment plus anti-
mycobacterial treatment group.
Of 268 patients with BUD with at least 1 positive laboratory
test result, 229 (85.4%) had a positive swab sample and/or tissue
specimen with a positive PCR result, and 152 (56.7%) had a
positive swab sample and/or a tissue specimen with positive
microscopy findings. One hundred forty-nine (98.0%) of the
152 specimens with positive microscopy findings had those
findings confirmed by at least 1 of the 3 other tests.
Of the 115 isolates (42.9%) obtained from swab sample and/
or tissue specimen cultures with positive results, 108 isolates
were confirmed by IS2404 PCR (positive predictive value,
93.9%). Among the remaining 7 isolates, sequence analysis
identified 2 M. ulcerans strains. Two further strains were iden-
tified as Mycobacterium mucogenicum and Mycobacterium pho-
caicum, indicating a coinfection or superinfection in 2 indi-
Different Diagnostic Methods for BUD • CID 2009:48 (15 April) • 1059
Table 1. Location of nonulcerative and ulcerative lesions in248 cases of laboratory-confirmed Buruli ulcer disease.
Specific location
No. (%) of lesions, by general location(n p 248)
Headand trunk
Right shoulderand limbs
Left shoulderand limbs Total
Head, neck 6 … … 6 (2.4)
Back 7 … … 7 (2.8)
Abdomen 0 … … 0 (0)
Buttock, hip 4 … … 4 (1.6)
Shoulders … 7 6 13 (5.2)
Arms
Overall … 65 39 104 (41.9)
Upper arm … 24 17 41 (16.5)
Forearm … 34 18 52 (21.0)
Wrist … 3 1 4 (1.6)
Hand, dorsal … 3 3 6 (2.4)
Hand, volar … 1 0 1 (0.4)
Legs
Overall … 66 48 114 (46.0)
Thigh … 20 13 33 (13.3)
Knee … 9 7 16 (6.5)
Lower leg … 26 18 44 (17.7)
Ankle … 8 8 16 (6.5)
Foot, dorsal … 3 2 5 (2.0)
Foot, plantar … 0 0 0 (0)
Overall 17 (6.9) 138 (55.6) 93 (37.5) 248 (100)
NOTE. For 20 cases, the specific location of the lesions was not known.
Figure 2. Age distribution of 268 patients with laboratory-confirmed Buruli ulcer disease (BUD). For 3 patients, age was unknown. Age range was2–80 years, the mean age was 18 years, and the median age was 12 years. For the group of patients who received drug treatment, patients withsuspected BUD who were !5 years of age were not included. For the surgical treatment without previous antimycobacterial treatment group and thesurgical treatment with previous antimycobacterial treatment group, patients with suspected BUD were included regardless of age.
viduals with confirmed BUD. For 3 isolates, sequencing did
not provide definitive identification, probably because of con-
tamination with closely related species that colonize the human
skin.
Among the follow-up samples, sequencing identified 2
strains as Mycobacterium gordonae (cultured from an ulcer with
a swab sample with positive IS2404 PCR) and Mycobacterium
szulgai (isolated from an additional lesion at a different location
on a patient whose initial lesion had a swab sample with positive
IS2404 PCR results).
Histopathological examination confirmed results for 42 of
49 tissue specimens from individuals with otherwise laboratory-
confirmed BUD (sensitivity, 85.7%) and for 17 (29.3%) of 58
specimens from individuals with suspected BUD who had neg-
ative microscopy findings and negative culture and PCR results
(6 [10.3%] of these 58 specimens were obtained from lesions
in the healing stages). In 4 (6.9%) of the 58 individuals with
suspected BUD, histopathological features did not allow an
unambiguous diagnosis, and histological findings in 37 (63.8%)
were not suggestive of BUD (8 of these 37 patients received a
diagnosis of onchocerciasis). Missing or poor-quality specimens
did not allow histopathological analysis for the remaining 75
individuals with suspected BUD.
Of 30 individuals who were followed up over time, 16 re-
ceived laboratory confirmation of BUD at their first presen-
tation to the hospital. Of the remaining 14 subjects, 4 had BUD
confirmed during subsequent follow-up visits. External quality
assurance for microscopic examination and PCR showed 190%
concordance of results (table 2).
Overall sensitivities of laboratory tests among all labora-
tory-confirmed BUD cases. The overall sensitivities were
85.4% (229 of 268 cases) for PCR, 56.7% (152 of 268 cases)
for microscopic examination, and 48.0% (108 of 225 cases) for
culture plus confirmatory IS2404 PCR. The sensitivity of PCR
was significantly higher than that of microscopic examination
and culture plus confirmatory IS2404 PCR ( ), with noP ! .01
statistically significant difference between microscopic exami-
1060 • CID 2009:48 (15 April) • Herbinger et al.
Table 2. External quality assurance for PCR and microscopic examination.
NOTE. Data are no. of specimens, unless otherwise indicated. DITM, Department of InfectiousDiseases and Tropical Medicine (Munich, Germany); KCCR, Kumasi Centre for Collaborative Researchin Tropical Medicine (Kumasi, Ghana). Microscopic examination was performed for 96 slides at KCCR,and the slides were reexamined by staff at DITM.
a No. of specimens with concordant results divided by the total number of specimens tested withthe same test at KCCR and DITM.
b PCR was performed with parallel testing of 77 DNA extracts. Dry reagent–based IS2404 PCRwas performed at KCCR; standard PCR was performed at DITM.
c Dry reagent–based PCR performed at KCCR gave false-positive results for 0 (0%) of the spec-imens tested and false-negative results for 7 (9.1%) of the specimens tested.
d Microscopic examination performed at KCCR gave false-positive results for 5 (5.2%) of thespecimens tested and false-negative results for 2 (2.1%) of the specimens tested.
nation and culture plus confirmatory IS2404 PCR ( )P p .054
(table 3).
Sensitivities of laboratory tests among confirmed BUD cases
involving nonulcerative lesions. Among 114 confirmed non-
treatment group, 15 cases; surgical treatment plus antimyco-
bacterial treatment group, 7 cases), the sensitivities were 89.5%
(102 of 114 specimens) for PCR, 57.0% (65 of 114 specimens)
for microscopic examination, and 60.6% (57 of 94 specimens)
for culture plus confirmatory IS2404 PCR analysis of tissue
specimens. PCR was significantly more sensitive than micro-
scopic examination and culture plus confirmatory IS2404 PCR
( for each). Stratified into treatment groups and speci-P ! .01
mens, the sensitivities for analysis of punch biopsy tissue spec-
imens from the drug treatment group were 93.5% for PCR and
70.8% for culture plus confirmatory IS2404 PCR, which was
statistically significantly higher than that for surgically excised
tissue specimens in the surgical treatment group (table 3).
Sensitivities of laboratory tests among patients with con-
firmed BUD and ulcerative lesions. Among 154 laboratory-
confirmed cases of ulcerative BUD (drug treatment group, 68
cases; surgical treatment group, 44 cases; surgical treatment plus
antimycobacterial therapy group, 42 cases), the sensitivities
were 69.8% (169 of 242 specimens) for PCR, 44.6% (108 of
242 specimens) for microscopic examination, and 25.0% (52
of 208 specimens) for culture plus confirmatory IS2404 PCR
analysis of swab samples and tissue specimens. PCR was sta-
tistically significantly more sensitive than microscopic exami-
nation and culture plus confirmatory IS2404 PCR (P ! .01
each). Stratified into treatment groups and specimens, the sen-
sitivities for PCR analysis of swab specimens were 89.9% for
the drug treatment group, 73.1% for the surgical treatment
group, and 72.2% for the surgical treatment plus antimyco-
bacterial therapy group. In all treatment groups, PCR sensitivity
was greater for swab samples than it was for punch biopsy
tissue specimens (67.8%) or surgically excised tissue specimens
(surgical treatment group, 57.7%; surgical treatment plus an-
timycobacterial therapy group, 44.4%) (table 3).
Sensitivities of laboratory tests among patients with ulcer-
ative lesions without previous antimycobacterial treatment de-
pending on the duration of disease. According to duration
of disease, 101 patients with laboratory-confirmed, previously
untreated cases of BUD with ulcerative lesions were divided
into 5 groups. In all groups, PCR of swab samples had a sta-
tistically significantly higher sensitivity (70%–91%) than did
microscopic examination of swab samples (30%–67%) or cul-
ture plus confirmatory IS2404 PCR of swab samples (29%–
59%). Despite a slight downward trend, no statistically signif-
icant association between test sensitivity and duration of disease
for any test was found (figure 3).
Test sensitivity depending on duration of previous anti-
mycobacterial treatment. In the 49 laboratory-confirmed
cases of BUD in the surgical treatment plus antimycobacterial
treatment group (7 nonulcerative case and 42 ulcerative cases),
the test sensitivity was correlated with the duration of previous
antimycobacterial treatment. Regardless of treatment duration,
PCR was always the test with the highest sensitivity. The sen-
sitivities of PCR and culture plus confirmatory IS2404 PCR
were statistically significantly higher in the group of untreated
patients than in the group of patients treated for 140 days (for
Table 3. Sensitivity of 3 different methods for the diagnosis of nonulcerative and ulcerative lesions of Buruli ulcer disease (BUD), by method of specimen collectionand treatment group.
Type of lesion, treatment group (specimen type)
Sensitivity, % (no. of positive results/no.of specimens tested) Pa
NOTE. Specimens were obtained for a total of 268 laboratory-confirmed cases of BUD. In 43 cases, however, no results for culture and confirmatory dry reagent–based IS2404 PCR were available.Therefore, the denominator for CUP is 225, not 268. CUP, culture and confirmatory PCR; DT, drug treatment with !7 days of previous antimycobacterial treatment; ST, surgical treatment with !7days of previous antimycobacterial treatment; ST+, surgical treatment with �7 days of previous antimycobacterial treatment.
a For comparison of sensitivity of tests performed on specimens taken from the same lesion.b Sensitivity of PCR was statistically significantly ( ) higher than that of microscopic examination and CUP.P ! .05c P value comparing the sensitivity of PCR, microscopic examination, and CUP between punch biopsy tissue specimens and surgically excised tissue specimens obtained from patients with
nonulcerative lesions.d Sensitivity of PCR was not statistically significantly ( ) higher than that of microscopic examination and was statistically significantly ( ) higher than that of CUP.P 1 .05 P ! .05e Sensitivity of microscopic examination was statistically significantly ( ) higher than that of CUP.P ! .05f Data for patients with BUD who had ulcerative lesions were included in the calculation for the swab samples if test results were obtained for both swab samples and tissue specimens taken
from the same lesion.g P value comparing the sensitivity of PCR, microscopic examination, and CUP between swab samples and tissue specimens obtained from the same ulcerative lesion.h P value comparing the sensitivity of PCR, microscopic examination, and CUP between punch biopsy tissue specimens and surgically excised tissue specimens obtained from the same ulcerative
lesion.i Includes all results, regardless of whether 1 specimens (swab sample or tissue specimen) or 2 specimens (swab sample and tissue specimen) were obtained per lesion.j Includes both nonulcerative and ulcerative lesions. For ulcerative lesions, only 1 result was included; if 2 results from the same ulcerative lesion (from a swab sample and a tissue specimen)
were not concordant, then the lesion was considered to be positive if 1 of the 2 tests had a positive result.
1062 • CID 2009:48 (15 April) • Herbinger et al.
Figure 3. Sensitivity of diagnostic tests (dry reagent–based [DRB] IS2404 PCR, microscopic examination, and culture plus confirmatory DRB IS2404PCR) by duration of disease among 112 patients with laboratory-confirmed Buruli ulcer disease and ulcerative lesions who had not received previousantibiotic treatment (68 swab specimens from patients in the drug treatment group and 44 swab specimens from patients in the surgical treatmentwithout prior antimycobacterial therapy group). No patient reported a 7–9-month duration of disease.
PCR) or 120 days (for culture plus confirmatory IS2404 PCR).
No statistically significant difference was found for microscopic
examination (figure 4).
DISCUSSION
This is, to our knowledge, the largest study to date to evaluate
the diagnostic yield of various diagnostic tests after the intro-
duction of antimycobacterial therapy in a West African region
in which BUD is highly endemic. Diagnostic tests proved to
be reliable within a range of disease durations and across a
range of clinical presentations and treatment groups. The results
are relevant for case definition for drug trials. Furthermore,
locally conducted laboratory confirmation may improve the
reliability of epidemiological data.
The majority of recent data on the laboratory confirmation
of clinically suspected BUD are from the era when surgical
excision was considered to be the standard treatment, when
surgically excised tissue specimens accounted for the majority
of specimens analyzed. Following the introduction of anti-
mycobacterial treatment, swab samples and punch biopsy tissue
specimens, which provide the possibility of pretreatment di-
agnosis, have become increasingly important. Surgical excision
and subsequent skin grafting, however, are still used to treat
patients who experience treatment failure and have lesions that
do not heal completely after antimycobacterial therapy. In these
cases, surgically excised tissue specimens are available for lab-
oratory analysis.
Comparable with previous data [20], laboratory confirma-
tion by 1 positive test result gave 26% more confirmed cases
than did confirmation by at least 2 positive test results. As in
previous studies, histopathological examination identified an
additional ∼30% of cases, mainly in patients whose disease was
in the paucibacillary stage [14, 19, 20]. However, histopatho-
logical features may not provide unambiguous identification,
and the availability of the method is limited [11, 14, 19, 20].
Findings obtained with follow-up samples suggest retesting of
patients who have typical clinical features but initial laboratory
results that are negative.
The overall sensitivities of dry reagent–based PCR (85%),
microscopic examination (57%), and culture plus confirmatory
IS2404 PCR (48%), as determined in our study, are comparable
with data published by other groups [8–15]. Independently of
treatment group, type of lesion, or diagnostic specimen, the
overall sensitivity of PCR was statistically significantly (P !
) higher than that of any other test, whereas there was no.01
statistically significant difference between microscopic exami-
nation and culture plus confirmatory IS2404 PCR ( ).P p .054
The majority of mycobacterial isolates were confirmed to be
M. ulcerans by IS2404 PCR or other methods. However, a few
study patients, most of whom had received antimycobacterial
treatment, harbored other mycobacteria. Because these strains
were only isolated from these patients, laboratory contamina-
tion is unlikely. In the absence of other confirmatory tests or
persistent mycobacterial growth after drug treatment, confir-
mation by molecular methods is important to identify coin-
fections or superinfections due to other mycobacteria.
Stratification by lesion type and treatment group gave the
following major findings. The sensitivity (93.5%) of PCR per-
formed on 3-mm punch biopsy tissue specimens obtained from
previously untreated patients in the drug treatment group who
Different Diagnostic Methods for BUD • CID 2009:48 (15 April) • 1063
Figure 4. Sensitivity of diagnostic tests (dry reagent–based [DRB] IS2404 PCR, microscopy, and culture plus confirmatory DRB IS2404 PCR) among49 patients with laboratory-confirmed cases of Buruli ulcer disease who received surgical treatment with previous antimycobacterial treatment (7 withpreulcerative and 42 with ulcerative lesions), stratified by duration of previous antimycobacterial treatment (3–20 days, 9 patients; 21–40 days, 30patients; �40 days, 10 patients). A total of 219 patients received no treatment.
had nonulcerative lesions was significantly higher than the sen-
sitivity of any other diagnostic test. These data are in line with
the 98.3% sensitivity of PCR of 4-mm and 6-mm punch biopsy
tissue specimens determined by Phillips et al. [23]. Culture plus
confirmatory IS2404 PCR of punch biopsy tissue specimens
also provided an excellent sensitivity of 70.8% among patients
with BUD who had nonulcerative lesions. Therefore, 3-mm
punch biopsy tissue specimens can be recommended for the
pretreatment diagnosis of patients with nonulcerative lesions
before initiation of antimycobacterial therapy. However, the
small size of punch biopsy tissue specimens hampers histo-
pathological analysis [29].
In previous studies in Ghana, PCR of swab samples con-
firmed 60%–70% of suspected cases of BUD with ulcerative
lesions [14, 20]. The data on the sensitivity of PCR of swab
samples obtained in this study also suggest considering PCR
analysis of diagnostic swab samples to be the method of choice
for cases that involve ulcerative lesions. Especially among the
previously untreated patients in the drug treatment group with
ulcerative early lesions, PCR of swab samples (sensitivity,
89.9%) proved to be superior to PCR analysis of 3-mm punch
biopsy tissue specimens (sensitivity, 67.8%). In the surgical
treatment group and the surgical treatment plus antimycobac-
terial treatment group, PCR of swab samples also had higher
sensitivity than did analysis of tissue specimens. In accordance
with previous observations, the lower sensitivity for tissue spec-
imens obtained from patients who underwent surgery may be
attributable to difficulties in determining the correct location
for specimen collection once tissue is excised [14].
Despite a slight downward trend, no statistically significant
association between duration of disease and diagnositic sen-
sitivity was detected for swab specimens in this study. However,
according to our own experience, in the course of the disease,
the edges of ulcers often develop scarring, which can hinder
the collection of swab samples. The duration of antimycobac-
terial treatment influenced the diagnostic sensitivities of PCR
and culture plus confirmatory IS2404 PCR. Compared with
their sensitivity in untreated patients, the sensitivities of both
tests were statistically significantly lower after treatment. A sta-
tistically significant decrease in culture sensitivity was detected
after 20 days of treatment. More than 40 days of treatment
were required to produce the same effect for PCR. In pretreated
patients, the sensitivity of PCR was still 50%, whereas the sen-
sitivity of culture plus confirmatory IS2404 PCR decreased to
6%. In contrast with the rapid decrease in the viability of M.
ulcerans in the first weeks after onset of treatment, PCR findings
suggest extended persistence of M. ulcerans DNA in treated
lesions.
This study describes the relative sensitivity of currently avail-
able diagnostic tests. Data on the specificity and the positive
predictive values of these tests cannot be provided, because
analysis of diagnostic samples from healthy individuals (in-
cluding tissue specimens) would have been required to deter-
mine the number of false-positive test results. Assessment of
positive and negative predictive values requires comparison
with a reference test. Because of the limited availability of ref-
erence methods (e.g., histopathological examination as a ref-
erence test for PCR), the determination of positive and negative
predictive values was not feasible for the entire range of tests
used in this study.
According to the results of this study, IS2404 PCR was the
test with the highest sensitivity overall and in all subgroups of
1064 • CID 2009:48 (15 April) • Herbinger et al.
this study cohort; therefore, it is most suitable for the early
diagnosis of all clinical forms of BUD. For monitoring of an-
timycobacterial treatment success within controlled trials, how-
ever, only culture seems to be the appropriate tool.
Acknowledgments
Financial support. European Commission (project no. INCO-CT-2005-051476-BURULICO).
Potential conflicts of interest. All authors: no conflicts.
References
1. Amofah G, Bonsu F, Tetteh C, et al. Buruli ulcer in Ghana: results ofa national case search. Emerg Infect Dis 2002; 8:167–70.
2. Sizaire V, Nackers F, Comte E, Portaels F. Mycobacterium ulcerans in-fection: control, diagnosis, and treatment. Lancet Infect Dis 2006; 6:288–96.
3. World Health Organization. Buruli ulcer progress report, 2004–2008.Weekly Epidemiological Record 2008; 83:145–56.
4. Buntine J, Crofts K, eds. Buruli ulcer: management of Mycobacteriumulcerans disease: a manual for health care providers. Geneva: WorldHealth Organization, 2001.
5. World Health Organization (WHO). Provisional guidance on the roleof specific antibiotics in the management of Mycobacterium ulceransdisease (Buruli ulcer). Geneva: WHO, 2004
6. Etuaful S, Carbonnell B, Grosset J, et al. Efficacy of the combinationrifampin-streptomycin in preventing growth of Mycobacterium ulceransin early lesions of Buruli ulcer in humans. Antimicrob Agents Che-mother 2005; 49:3182–6.
7. Chauty A, Ardant MF, Adeye A, et al. Promising clinical efficacy ofstreptomycin-rifampin combination for treatment of buruli ulcer (My-cobacterium ulcerans disease). Antimicrob Agents Chemother 2007; 51:4029–35.
8. Asiedu K, Wansbrough-Jones M. Mycobacterium ulcerans infection(Buruli or Bairnsdale ulcer): challenges in developing managementstrategies. Med J Aust 2007; 186:55–6.
9. Portaels F, Agular J, Fissette K, et al. Direct detection and identificationof Mycobacterium ulcerans in clinical specimens by PCR and oligo-nucleotide-specific capture plate hybridization. J Clin Microbiol1997; 35:1097–100.
10. Guimaraes-Peres A, Portaels F, de Rijk P, et al. Comparison of twoPCRs for detection of Mycobacterium ulcerans. J Clin Microbiol1999; 37:206–8.
11. Guarner J, Bartlett J, Whitney EA, et al. Histopathologic features ofMycobacterium ulcerans infection. Emerg Infect Dis 2003; 9:651–6.
12. Stienstra Y, van der Werf TS, Guarner J, et al. Analysis of an IS2404-based nested PCR for diagnosis of Buruli ulcer disease in regions ofGhana where the disease is endemic. J Clin Microbiol 2003; 41:794–7.
13. Yeboah-Manu D, Bodmer T, Mensah-Quainoo E, Owusu S, Ofori-AdjeiD, Pluschke G. Evaluation of decontamination methods and growthmedia for primary isolation of Mycobacterium ulcerans from surgicalspecimens. J Clin Microbiol 2004; 42:5875–6.
14. Siegmund V, Adjei O, Nitschke J, et al. Dry reagent–based polymerasechain reaction compared with other laboratory methods available forthe diagnosis of Buruli ulcer disease. Clin Infect Dis 2007; 45:68–75.
15. Mensah-Quainoo E, Yeboah-Manu D, Asebi C, et al. Diagnosis ofMycobacterium ulcerans infection (Buruli ulcer) at a treatment centrein Ghana: a retrospective analysis of laboratory results of clinicallydiagnosed cases. Trop Med Int Health 2008; 13:191–8.
16. Ross BC, Marino L, Oppedisano F, Edwards R, Robins-Browne RM,Johnson PD. Development of a PCR assay for rapid diagnosis of My-cobacterium ulcerans infection. J Clin Microbiol 1997; 35:1696–700.
17. Stinear T, Ross BC, Davies JK, et al. Identification and characterizationof IS2404 and IS2606: two distinct repeated sequences for detectionof Mycobacterium ulcerans by PCR. J Clin Microbiol 1999; 37:1018–23.
18. Rondini S, Mensah-Quainoo E, Troll H, Bodmer T, Pluschke G. De-velopment and application of real-time PCR assay for quantificationof Mycobacterium ulcerans DNA. J Clin Microbiol 2003; 41:4231–7.
19. Siegmund V, Adjei O, Racz P, et al. Dry reagent-based PCR as a noveltool for laboratory confirmation of clinically diagnosed Mycobacteriumulcerans–associated disease in areas in the tropics where M. ulcerans isendemic. J Clin Microbiol 2005; 43:271–6.
20. Bretzel G, Siegmund V, Nitschke J, et al. A stepwise approach to thelaboratory diagnosis of Buruli ulcer disease. Trop Med Int Health2007; 12:89–96.
21. Johnson PD, Hayman JA, Quek TY, et al.; Mycobacterium ulceransStudy Team. Consensus recommendations for the diagnosis, treatmentand control of Mycobacterium ulcerans infection (Bairnsdale or Buruliulcer) in Victoria, Australia. Med J Aust 2007; 186:64–8.
22. Portaels F, Johnson P, eds. Buruli ulcer: diagnosis of Mycobacteriumulcerans disease: a manual for health care providers. Geneva: WorldHealth Organization, 2001.
23. Phillips R, Horsfield C, Kuijper S, et al. Sensitivity of PCR targetingthe IS2404 insertion sequence of Mycobacterium ulcerans in an assayusing punch biopsy specimens for diagnosis of Buruli ulcer. J ClinMicrobiol 2005; 43:3650–6.
24. Bretzel G, Siegmund V, Nitschke J, et al. External quality assurance forthe laboratory diagnosis of Buruli ulcer disease in Ghana. Trop MedInt Health 2006; 11:1688–93.
25. Kim BJ, Lee SH, Lyu MA, et al. Identification of mycobacterial speciesby comparative sequence analysis of the RNA polymerase gene (rpoB).J Clin Microbiol 1999; 37:1714–20.
26. Talaat AM, Reimschuessel R, Trucksis M. Identification of mycobacteriainfecting fish to the species level using polymerase chain reaction andrestriction enzyme analysis. Vet Microbiol 1997; 58:229–37.
27. Roth A, Reischl U, Streubel A, et al. Novel diagnostic algorithm foridentification of mycobacteria using genus-specific amplification of the16S–23S rRNA gene spacer and restriction endonucleases. J Clin Mi-crobiol 2000; 38:1094–104.
28. Kim H, Kim SH, Shim TS, et al. PCR restriction fragment lengthpolymorphism analysis (PRA)-algorithm targeting 644 bp heat shockprotein 65 (hsp65) gene for differentiation of Mycobacterium spp. JMicrobiol Methods 2005; 62:199–209.
29. Herbinger KH, Brieske D, Nitschke J, et al. Excision of pre-ulcerativeforms of Buruli ulcer disease: a curative treatment? Infection 2009; 37:20–5.
Post-surgical assessment of excised tissue from patients with
Buruli ulcer disease: progression of infection in macroscopically
healthy tissue
G. Bretzel1,2, V. Siegmund1,2, P. Racz1, F. van Vloten1, F. Ngos3, W. Thompson4, P. Biason5, O. Adjei6,
B. Fleischer1 and J. Nitschke1,5
1 Bernhard Nocht Institute for Tropical Medicine (BNITM), Hamburg, Germany2 Department of Infectious Diseases and Tropical Medicine, University of Munich, Germany3 Hopital de District, District de Sante d’Akonolinga, Akonolinga, Cameroon4 Agogo Presbyterian Hospital, Agogo, Ghana5 Medecins sans Frontieres (MSF), Geneva, Switzerland6 Kumasi Centre for Collaborative Research in Tropical Medicine (KCCR), University of Science and Technology, Kumasi, Ghana
Summary objective The current standard of treatment of Buruli ulcer disease (BUD) is surgical excision of
lesions. Excision size is determined macroscopically assuming the complete removal of all infected tissue.
However, dissemination of infection beyond the excision margins into apparently healthy tissue,
possibly associated with recurrences, cannot be excluded in this way. To assess the central to peripheral
progression of Mycobacterium ulcerans infection and the mycobacterial infiltration of excision margins,
excised tissue was examined for signs of infection.
methods 20 BUD lesions were excised in general anaesthesia including all necrotic and subcuta-
neous adipose tissue down to the fascia and at an average of 40 mm into the macroscopically unaffected
tissue beyond the border of the lesion. Tissue samples were subjected to PCR and histopathology.
results Although the bacillary load decreased from central to peripheral, M. ulcerans infection
was detected throughout all examined tissue specimens including the peripheral segments as well as
excision margins of all patients. During the post-operative hospitalization period (averaging 2 months)
no local recurrences were observed.
conclusion Available data suggest a correlation of surgical techniques with local recurrences. The
results of this study indicate the unnoticed early progression of mycobacterial infection into macro-
scopically healthy tissue. Thus, the removal of all infected tissue cannot always be verified visually by
the surgeon. Provided that long-term follow up of patients with positive excision margins will establish
the clinical relevance of these findings, on-site laboratory assessment of excised tissue in combination
with follow up may contribute to reduce recurrence rates.
P. Racz, F. van Vloten and B. Fleischer, Bernhard Nocht Institute for Tropical Medicine (BNITM), Hamburg, Germany.
F. Ngos, Hopital de District, District de Sante d’Akonolinga, Akonolinga, Cameroon.
W. Thompson, Agogo Presbyterian Hospital, Agogo, Ghana.
P. Biason, Medecins sans Frontieres (MSF), Geneva, Switzerland.
O. Adjei, Kumasi Centre for Collaborative Research in Tropical Medicine (KCCR), University of Science and Technology, Kumasi,
Ghana.
J. Nitschke, Bernhard Nocht Institute for Tropical Medicine (BNITM), Hamburg, Germany; Medecins Sans Frontieres (MSF), Geneva,
Switzerland.
Tropical Medicine and International Health volume 10 no 11 pp 1199–1206 november 2005
G. Bretzel et al. Post-surgical assessment of Buruli ulcer tissue
1206 ª 2005 Blackwell Publishing Ltd
Excision of Pre-Ulcerative Forms of Buruli UlcerDisease: A Curative Treatment?
K.-H. Herbinger, D. Brieske, J. Nitschke, V. Siegmund, W. Thompson, E. Klutse,N.Y. Awua-Boateng, E. Bruhl, L. Kunaa, M. Schunk, O. Adjei, T. Löscher, G. Bretzel
AbstractBackground: Previous investigations have revealed thatMycobacterium ulcerans is extensively distributed spatiallythroughout ulcerative lesions, including in the margins ofexcised tissue. In contrast, bacilli in pre-ulcerative lesions areassumed to be concentrated in the center of the lesion. Inorder to assess the extent to which the surgical excision ofpre-ulcerative lesions is capable of removing all infectedtissue, we subjected the excision margins of pre-ulcerativelesions to laboratory analysis.Patients and Methods: Eleven patients with laboratory-confirmed pre-ulcerative lesions were included in thestudy. The diameter of the lesion and excised tissue andthe ‘‘surgical distance’’ between the border of the lesionand excision margin were measured. The entire excisionmargin was cut into segments and subjected to IS2404PCR.Results: The results from the PCR analysis on the samplesof excision margins were highly significantly associatedwith the surgical distance (p < 0.001). The margin samplesof nodules were significantly more often PCR positive thanthe plaques (p = 0.025). The size of the lesion and the sizeof the excised tissue did not significantly influence thePCR results. Statistically, a surgical distance of more than9 mm was found to reduce the risk of remaining infectedtissue to less than 10%, that of 13 mm to reduce the riskto less than 5%, and that of 25 mm to reduce the risk tonearly 0%.Conclusion: The results of this study show that in pre-ulcerative Buruli ulcer disease, bacilli may extend beyondthe actual size of the lesion and that there is a strongcorrelation between the presence of M. ulcerans in themargin samples and the surgical distance. Excision with asurgical distance of 25 mm avoided the risk of remainingmycobacteria in this study. However, no recurrencesoccurred in the patients with M. ulcerans-positiveexcision margins. The need of postoperative antimycobac-terial treatment in these patients remains to bedetermined.
Infection 2008
DOI 10.1007/s15010-008-8073-4
IntroductionBuruli ulcer disease (BUD) caused by Mycobacteriumulcerans is an infectious disease involving the skin and thesubcutaneous adipose tissue. It has been reported or sus-pected in 30 countries worldwide, with a focus andincreasing case numbers in West Africa [1–3]. The diseaseinitially presents as painless nodule, papule, plaque, oredema (pre-ulcerative forms), evolving into a painlessulcer with characteristically undermined edges (ulcerativeform). While mortality is low, serious long-term sequelaeare common. Contractures and deformities originatingfrom self-healing processes or surgical treatment of late,extensive lesions may result in severe functional defi-ciencies [4, 5].
Various distributions of pre-ulcerative vs ulcerativeforms have been reported. In Cote d’Ivoire, Kanga et al.observed only 4% of nodular and 6.5% of edematousforms vs 89.5% of ulcerated cases [6]. In Benin, the fre-quency of pre-ulcerative cases was reported to be 37.4%(7.6% nodules, 29.1% plaques, 0.7% edema) [3]. InGhana, 42.6% of the clinically diagnosed BUD patientspresented with nodules (31.3%), plaques (10.9%), orpapules (0.4%) [7].
Due to the characteristically clinical presentation ofulcerative lesions, experienced clinicians are generally
K.-H. Herbinger (corresponding author), D. Brieske, V. Siegmund,M. Schunk, T. Löscher, G. BretzelDept. of Infectious Diseases and Tropical Medicine (DITM), Ludwig-Maximilian University Munich, Leopoldstrasse 5, 80802 Munich,Germany; Phone: (+49/89) 2180-3618, Fax: -336112,e-mail: [email protected]. NitschkeBernhard Nocht Institute for Tropical Medicine (BNITM), Hamburg,GermanyW. ThompsonAgogo Presbyterian Hospital, Agogo, GhanaE. KlutseDunkwa Governmental Hospital, Dunkwa-on-Offin, GhanaN.Y. Awua-Boateng, E. Bruhl, L. Kunaa, O. AdjeiKumasi Centre for Collaborative Research in Tropical Medicine (KCCR),Kumasi, Ghana
Received: February 21, 2008 Æ Revision accepted: June 24, 2008Published online: December 3, 2008
Infection Clinical and Epidemiological Study
Infection 2008 � URBAN & VOGEL
able to establish the diagnosis on clinical grounds, with theanalysis of diagnostic swabs providing laboratory confir-mation in the majority of cases. In contrast, pre-ulcerativelesions have uncharacteristic clinical features and, conse-quently, a variety of possible differential diagnoses exists.An accurate clinical diagnosis is therefore more difficult,and priority should be given to the laboratory confirma-tion or differential diagnosis of these cases based on ananalysis of tissue specimens [7–9].
The effect of antibiotic treatment alone or in com-bination with surgery has recently come under investiga-tion. Data from Ghana, Benin, and Australia suggest thatthere is a benefit to treating pre-ulcerative and ulcerativeforms of BUD with antibiotics [9–13]. A small Australiancase series of early ulcerative lesions lead to the formu-lation of a treatment algorithm for ulcerative lesions andcellulitis that includes perioperative antimycobacterialtreatment, surgical debridement with 3- to 5- mm margins(to be repeated until the margins are microscopicallyclear of acid-fast bacilli (AFB), followed by at least2 months of antimycobacterial treatment [14]. Dependingon the size and type of the lesion, current provisionalWorld Health Organization guidelines recommend com-bination therapy with rifampicin and streptomycin for 4–8weeks [9–13].
Previous investigations of the spatial distribution ofM. ulcerans in ulcerative lesions have revealed the pres-ence of bacilli throughout the lesions as well as in mac-roscopically healthy excision margins [15, 16]. This findingindicates that even wide surgical excision of ulcerativelesions cannot guarantee the complete removal of all in-fected tissue. Remaining bacilli, however small theamount, are closely related to the possibility of recur-rences, a finding which supports the need for antimyco-bacterial treatment in ulcerative lesions. For nodularforms, however, bacilli have been shown to be concen-trated in the center of the lesion [17]. Therefore, excisionof pre-ulcerative forms with sufficiently wide excisionmargins may provide for the removal of all infected tissue.Data and recommendations on the management of pre-ulcerative lesions are scarce [5, 18, 19]. Observations
made during a follow-up study in Ghana (own unpub-lished data) suggest that functional limitations rarelyoccur following the excision of nodules and plaques,whereas surgical treatment of ulcerative lesions often re-sults in disabilities caused by extensive scarring associatedwith grafting. In-patient time was found to be significantlyshorter for nodules and plaques than for ulcerativelesions.
As several reasons may favor the surgical treatmentof pre-ulcerative lesions, this study was undertaken toinvestigate if, and under which conditions, primary sur-gery of nodules and plaques alone, i.e. without previousantimycobacterial treatment, may be curative.
Patients and MethodsPatients, Specimens, and Laboratory Analysis
Eleven patients with laboratory-confirmed pre-ulcerative lesions(nodules n = 7, plaques n = 4) without previous antimycobacte-rial treatment were eligible for this study. The patients under-went routine surgical excision in Agogo Presbyterian Hospital,Agogo, Ghana (nodules n = 3, plaques n = 3), and DunkwaGovernmental Hospital, Dunkwa-on-Offin, Ghana (nodulesn = 4, plaques n = 1) in 2006. As is the case for specimens forroutine laboratory confirmation, the samples used for analysis ofthe excision margins were taken from surgically excised tissue.Due to the introduction of antimycobacterial treatment in Ghanain 2006, only a limited number of pre-ulcerative lesions frompatients treated by surgery only could be analyzed. Informationon the treatment outcome was obtained by follow-up of the studypatients 12–18 months after the surgical intervention.
Ethical clearance for the study was provided by the Com-mittee of Human Research Publication and Ethics at the Schoolof Medical Sciences at the University of Science and Technology,Kumasi, Ghana. Informed patient consent was obtained frompatients before surgery.
Before surgery, the excision line comprising an area ofmacroscopically unaffected tissue was determined and markedby a sterile pen, and the medio-lateral and proximo-distaldiameters of the lesion and excision were measured in situ withdisposable paper rulers (Figure 1a). Excisions, including allsubcutaneous adipose tissue down to the fascia, were per-formed. The excised tissue was spread out on a table on sterilecotton gauze. Based on the anatomical location, which hadbeen determined prior to the surgery, the position of the ex-
Figure 1. a) Marking of excisionline and in situ measurement ofproximo-distal and medio-lateraldiameters of lesion and excision.b) Measurement of proximo-distal(12 h–6 h) and medio-lateral(3 h–9 h) diameters of lesion andexcision in excised tissue accordingto the previously determinedanatomical position of the excisedtissue. Surgical distance: distancebetween the border of the lesionand the margin of excised tissue(white arrows).
K.-H. Herbinger et al. Excision of Pre-Ulcerative Forms of Buruli Ulcer Disease
Infection 2008 � URBAN & VOGEL
cised tissue was marked in a clockwise direction (proximal:12 h, distal: 6 h). Due to possible post-surgical changes in theproportions of the excised tissue relative to the in situ mea-surements, the proximo-distal (12 h–6 h) and medio-lateral(9 h–3 h) diameters of the lesion as well as of the total excisionwere measured again. These measurements served as the basisfor further analysis. The distance between the macroscopicallyvisible border of the lesion and the outer excision margin isreferred to as the ‘‘surgical distance’’ (Figure 1b). In order toobtain as accurate measurements of the lesion as possible, thebottom side of the tissue was also examined for better identi-fication of the necrotic area.
The entire excision margin of each lesion was cut off(Figure 2a) and divided into segments with a sterile scalpel.Depending on the size of the lesion and excision, the size of themargin samples could slightly differ among the patients, with amaximum size of 10 mm · 10 mm. For each margin segment, thesurgical distance was measured individually (Figure 2b). Diag-nostic specimens were taken from the center of the lesionaccording to standardized criteria and subjected to microscopy,culture, IS2404 PCR, and histopathology according to standardprocedures [7, 8]. A total of 150 margin segments were collected,stored in cell lysis solution (Gentra Systems, Minneapolis, MN)and tested for M. ulcerans by IS2404 PCR. Briefly, DNA fromtissue specimens was prepared using the Puregene DNA Isola-tion kit (Genomic DNA Purification kit, Gentra Systems), andthe IS2404 PCR was performed according to Stinear et al. [20]. Inorder to exclude false positive PCR results due to contamination,for each PCR extraction, negative controls were performed in astandardized manner [7]. Given the 100% specificity of theIS2404 PCR for clinical samples demonstrated in several studies[7, 8, 12], it was not considered necessary to confirm positivePCR results of margin samples by histopathology.
Statistical AnalysisAll analyses were performed using Stata software, ver. 9.0. (StataCorp., College Station, TX). For simple statistical analyses,approximative tests (v2 tests) and exact tests (Fisher’s exact tests)were applied. For calculating the association between thedependent variable (PCR results for the samples of the margins)and the independent variables, single and multiple logisticregression were used. In our analysis, we included the followingindependent variables: Hospital (Agogo, Dunkwa), sex and ageof patients, type of pre-ulcerative lesion (nodule, plaque), loca-tion of pre-ulcerative lesion, duration between first occurrence oflesion and excision, size of lesion, size of excised tissue, andsurgical distance.
ResultsBaseline Data
The baseline data of the study patients are summarized intable 1. Of 11 patients (nine females, two males) with pre-ulcerative lesions, seven had nodules and four had pla-ques. Six of them were operated on in Agogo and five inDunkwa. The youngest patient was 2 years old, and theoldest was 50 years old (arithmetic mean of age13.45 years).
The pre-ulcerative lesions were localized as follows: sixon thigh or lower leg, two on the arm, and one each on thehip, back, and foot. The duration between the first occur-rence of the lesion and excision was between 1 and 20 weeks(arithmetic mean of duration: 4.5 weeks). The smallest le-sion measured 47 mm2 (6 mm · 10 mm) and the largest707 mm2 (30 mm · 30 mm). The mean size of the lesionswas 279 mm2 (corresponding to approximately 19 mm ·19 mm). The smallest sample of excised tissue measured
176 mm2 (14 mm · 16 mm), and the largest measured4,712 mm2 (75 mm · 80 mm). The mean size of the excisedtissue was 1,783 mm2 (corresponding to approximately48 cm · 48 mm). The surgical distance was between 1 mmand 43 mm (arithmetic mean of surgical distance: 17 mm)depending on the anatomical conditions of the patient.
The number of margin samples collected per patientranged from 4 to 27, depending on the size of the pre-ulcerative lesion (arithmetic mean: 13.6 margin samplesper patient). This difference was corrected by multiplelogistic regression (all data of margin samples were ad-justed for patients).
All patients were considered to be healed when dis-charged from hospital. Our follow-up of the study patients12–18 months after the surgical intervention revealed norecurrences.
Association between PCR Results and IndependentVariables
The IS2404 PCR analysis obtained positive results for 20(13.3%) of the 150 margin samples. The association be-tween PCR results and independent variables is sum-marized in table 2. The detection of M. ulcerans
Figure 2. a) Abscission of excisionmargin. b) Cutting into marginsegments with a maximum sizeof 10 · 10 mm. The ‘‘surgicaldistance’’ was determined foreach margin segment individually.
K.-H. Herbinger et al. Excision of Pre-Ulcerative Forms of Buruli Ulcer Disease
Infection 2008 � URBAN & VOGEL
depended significantly on the surgical distance (the lar-ger the distance, the smaller the probability for themargin sample to be PCR positive; p < 0.001). Fractionalpolynomial regression showed that a surgical distance ofmore than 9 mm reduced the risk for any mycobacteriaremaining after excision to less than 10%, a surgicaldistance of more than 13 mm reduced the risk to lessthan 5%, and a surgical distance of more than 25 mmreduced the risk to nearly 0% (Figure 3). This regressionincludes all margin samples collected from nodules andplaques. Regression after stratification in the differenttypes of pre-ulcerative lesions showed x-axes crossing at26.8 mm for nodules and 22.9 mm for plaques. Patientswith nodules had a higher risk of being tested positivelythan patients with plaques (p = 0.025, adjusted for sur-gical distance).
The PCR results also depended on the age of thepatients. Older patients had a significantly smaller risk ofpositive margin samples (p = 0.014). This association wasconfounded by the surgical distance. The surgical distanceamong older patients was larger than that among youngerones (p = 0.007). The x-axes crossing was 29.3 mm forchildren (2–6 years), 19.2 mm for the adolescents (11–18 years), and 18.5 mm for the two adult patients (23 and50 years). After adjustment, the association between ageof the patients and the PCR result was not significant(p = 0.220, adjusted for surgical distance).
The independent variables hospital (Agogo Presby-terian Hospital, Dunkwa Government Hospital) and sexshowed a significant association with the PCR results(p = 0.014 and p = 0.025, respectively), but they were alsosignificantly associated with the variable surgical distance.
Table 2Association between PCR results of the margins of surgically excised tissue and different independent variables (without and withadjustment for surgical distance).
Number Independent variable p-value not adjusted p-value adjusteda
1 Hospital (Agogo, Dunkwa) 0.014* 0.1692 Age of patient 0.014* 0.2203 Sex of patient 0.025* 0.3044 Type of pre-ulcerative lesionb 0.006* 0.025*5 Location of pre-ulcerative lesion 0.183 0.1666 Durationc 0.122 0.1847 Size of pre-ulcerative lesion 0.078 0.1278 Size of excised tissue 0.243 0.6289 Surgical distance <0.001* –
*Significant p-values (< 0.050); a Adjusted for surgical distance; b Type of pre-ulcerative lesion of Buruli Ulcer: nodule or plaque; c Durationbetween first occurrence of lesion and excision
Table 1Patients with laboratory-confirmed pre-ulcerative lesions.
Numberof patients
Hospital Age(years)
Sexa Numberof samples
Typeof lesionb
Locationof lesion
Durationc
(weeks)Size of Ld
(mm2)Sizeof Ee (mm2)
Surgicaldistancef (mm)
PCRg
1 Dunkwa 4 F 9 N Arm 2 47 327 5–12 3/9 (33.3%)2 Dunkwa 11 F 12 P Leg 2 314 1,734 10–20 1/12 (8.3%)3 Dunkwa 6 F 11 N Leg 1 314 707 2–8 3/11 (27.3%)4 Agogo 12 M 17 N Leg 3 314 2,827 15–23 0/17 (0%)5 Agogo 23 F 15 P Arm 1 176 1,616 12–23 0/15 (0%)6 Agogo 2 M 24 P Back N.A. 311 3,792 21–29 0/24 (0%)7 Agogo 3 F 27 N Leg 8 314 4,712 17–43 2/27 (7.4%)8 Agogo 13 F 4 N Hip 8 79 176 1–5 2/4 (50%)9 Agogo 18 F 6 P Foot 20 174 307 2–3 1/6 (16.7%)
10 Dunkwa 50 F 17 N Leg 4 707 2,919 11–19 0/17 (0%)11 Dunkwa 6 F 8 N Leg 2 314 491 2–3 8/8 (100%)p-valueh 0.169 0.220 0.304 – 0.025* 0.166 0.184 0.127 0.628 < 0.001* –
* Significant p-values (< 0.050); N.A.: not available; a Sex: F, female; M, male; b Type of pre-ulcerative lesion of Buruli Ulcer: N, nodule;P, Plaque; c Duration between occurrence of lesion and excision; d L = Size of surface of pre-ulcerative lesion (nodule, plaque); e E = Size ofsurface of excised tissue; f Surgical distance: distance between the margin of the pre-ulcerative lesion and the margin of excised tissue;g Number of positive margin samples/number of all tested (positive and negative) margin samples. Tested by IS2404 PCR for M. ulcerans; h p-value of the association between the dependent variable (PCR result for M. ulcerans) and the independent variables, adjusted for surgicaldistance
K.-H. Herbinger et al. Excision of Pre-Ulcerative Forms of Buruli Ulcer Disease
Infection 2008 � URBAN & VOGEL
Multiple logistic regression was able detect this con-founding (hospital: p = 0.169, adjusted for surgical dis-tance; sex: p = 0.304, adjusted for surgical distance).
The independent variables location of lesion(p = 0.166), duration between first occurrence of lesionand excision (p = 0.184), size of lesion (p = 0.127), andsize of excision (p = 0.628) did not significantly influencethe PCR results of the margin samples.
DiscussionIn accordance with recent promising results, antibiotictreatment of BUD alone or in combination with surgerybecomes increasingly important [9–14]. The benefit oftreating ulcerative lesions with antibiotics is absolute andnot open to controversy. Even if antibiotic treatmentalone is not curative and subsequent surgery is required,antimycobacterial treatment can reduce the surface areaof the lesions, resulting in a less extensive excision. Inaddition, in view of the extensive spatial distribution ofthe bacilli and their spread into macroscopically healthyexcision margins, prior antimycobacterial treatment in-creases the chances of curative secondary surgery [15–17].
For pre-ulcerative lesions, however, other aspectsshould be taken into consideration. In a follow-up study inGhana (own unpublished data), the duration of the meanin-patient time after surgery for ulcerative lesions,84.0 days [95% confidence interval (CI) 65.3–102.7 days],exceeded the 8-week course of antibiotics. In contrast, themean in-patient time after surgical treatment of pre-ulcerative lesions, 33.4 days, was significantly shorter(95% CI 21.2–45.1 days; p = 0.002). No significant dif-ference in the mean in-patient time between nodules andplaques was found (p = 0.264). Primary surgery of theselesions therefore offers a less time-consuming treatment
option for the patient than antimycobacterial treatment.In the same follow-up study, functional limitations, suchas a reduced range of motion of joints after primary sur-gical excision, were not observed for patients with nodules(0/12) and only in a minority of patients with plaques (3/13: 23%). In this respect, there are no advantages ofantibiotic treatment over the surgical treatment of pre-ulcerative lesions.
Uncharacteristic clinical features and a variety ofpossible differential diagnoses of pre-ulcerative lesions(e.g., onchocercosis, leprosy, leishmaniasis, bacterial ab-scesses, lipoma) necessitate laboratory analysis to estab-lish the diagnosis. Punch biopsies or fine needle aspirates(FNA) enable a pre-surgical laboratory assessment oftissue specimens. However, the possibilities of establish-ing the differential diagnosis are limited. Fine needleaspirate specimens are not suitable for histopathologicalanalysis, and punch biopsies have been found to be toosmall to allow a reliable histopathological diagnosis inmany cases (personal communication, P. Racz, BernhardNocht Institute for Tropical Medicine, Hamburg, Ger-many and A. Schipf, Kantonsspital Luzern, Switzerland).Surgically excised tissue still provides the best option for areliable histopathological differential diagnosis of pre-ulcerative lesions, which also constitutes a valid argumentfor surgical excision in these cases [7–9].
The results of this study suggest a strong correlationbetween the surgical distance and the detection of M.ulcerans in margin samples by IS2404 PCR. Safe surgicaldistances of at least 25 mm theoretically reduce the riskof remaining infected tissue to 0%, whereas a surgicaldistance of at least 13 mm or 9 mm results in a risk ofless than 10 or 5%, respectively. We therefore concludethat primary surgery has the potential to be curative interms of pre-ulcerative lesions if the anatomical condi-tions allow excision with safe surgical distances. In viewof these results, surgical debridement with 3- to 5-mmmargins as recommended by Chin-Lenn et al. [14] doesnot seem to be sufficient for a curative excision of pre-ulcerative lesions. In the Australian case series, micro-scopic analysis was used to determine if the marginswere clear of AFB. However, the application of a moresensitive method, such as IS2404 PCR, which allows thedetection of small quantities of bacilli in microscopicallynegative tissue specimens, suggests the need for widersurgical distances.
Surgical excision of pre-ulcerative lesions can eitherbe carried out on an outpatient basis or with in-patienttimes significantly shorter than for those necessary forantimycobacterial therapy. Primary surgery can removemost of the infected tissue located in the center of thelesion, and it offers the possibility for reliable laboratorydiagnosis and differential diagnosis through the assess-ment of surgically excised tissue specimens.
In patients where anatomical conditions do not allowsufficiently wide excisions, a combination of primary
Figure 3. Regression (fractional polynomial): correlation betweenthe PCR results of the margin samples and the surgical distance(distance between the macroscopically visible border of pre-ulcerative lesion and the margin of excised tissue). Y-axes in logscale. Probability of a predicted positive PCR result at a surgicaldistance of more than 9.0 mm is below 10%, that at a surgicaldistance of more than 13 mm is below 5%, and that at a surgicaldistance of more than 25 mm is nearly 0%.
K.-H. Herbinger et al. Excision of Pre-Ulcerative Forms of Buruli Ulcer Disease
Infection 2008 � URBAN & VOGEL
surgery and subsequent antimycobacterial treatmentcould be considered. However, as no recurrences wereobserved in seven of our study patients with M. ulcerans-positive excision margins, the need for postoperative an-timycobacterial treatment in such patients remains to bedetermined. Further studies with a larger sample size arerequired to evaluate the benefit of a combination of pri-mary surgery and subsequent drug treatment in pre-ulcerative lesions.
AcknowledgmentsThis work was supported by the European Commission (projectINCO-CT-2005-051476-BURULICO).
Conflict of Interest Statement: The authors of this studydo not have a conflict of interest.
References
1. Amofah G, Bonsu F, Tetteh C, Okrah J, Asamoa K, Asiedu K,Addy J: Buruli ulcer in Ghana: results of a national case search.Emerg Infect Dis 2002; 8: 167–170.
2. World Health Organization (WHO) Buruli ulcer disease. Myco-bacterium ulcerans infection: an overview of reported casesglobally. Wkly Epidemiol Rec 2004; 79: 194–199.
3. Debacker M, Aguiar J, Steunou C, Zinsou C, Meyers WM,Guédénon A, Scott JT, Dramaix M, Portaels F: Mycobacteriumulcerans disease (Buruli ulcer) in rural hospital, Southern Benin1997–2001. Emerg Infect Dis. 2004; 10: 1391–1398.
4. World Health Organization (WHO) Buruli ulcer. Diagnosis ofMycobacterium ulcerans disease. WHO, Geneva, 2001.
5. World Health Organization (WHO) Buruli ulcer. Management ofMycobacterium ulcerans disease. WHO, Geneva, 2001.
6. Kanga JM, Kacou ED: Epidemiological aspects of Buruli ulcer inCôte d’Ivoire: results of a national survey. Bull Soc Pathol Exot2001; 94: 46–51.
7. Siegmund V, Adjei O, Nitschke J, Thompson W, Klutse E,Herbinger KH, Thompson R, van Vloten F, Racz P, Fleischer B,Loescher T, Bretzel G: Dry reagent-based polymerase chainreaction compared with other laboratory methods available forthe diagnosis of Buruli ulcer disease. Clin Infect Dis 2007; 45:68–75.
8. Bretzel G, Siegmund V, Nitschke J, Herbinger KH, Thompson W,Klutse E, Crofts K, Massavon W, Etuaful S, Thompson R,Asamoah-Opare K, Racz P, Vloten F, van Berberich C, Kruppa T,Ampadu E, Fleischer B, Adjei O: A stepwise approach to thelaboratory diagnosis of Buruli ulcer disease. Trop Med Int Health2007; 12: 89–96.
9. Chauty A, Ardant MF, Adeve A, Euverte H, Guédénon A, JohnsonC, Aubry J, Nuermberger E, Grosset J: Promising clinical efficacyof the combination streptomycin – rifampin for the treatmentof Buruli ulcer (Mycobacterium ulcerans disease). AntimicrobAgents Chemother 2007; 51: 4029–4035.
10. Etuaful S, Carbonnelle B, Grosset J, Lucas S, Horsfield C, PhillipsR, Evans M, Ofori-Adjei D, Klutse E, Owusu-Boateng J, AmedofuGK, Awuah P, Ampadu E, Amofah G, Asiedu K, Wansbrough-Jones M: Efficacy of the combination rifampicin-streptomycin inpreventing growth of Mycobacterium ulcerans in early lesions ofBuruli ulcer in humans. Antimicrob Agents Chemother 2005; 49:3182–3186.
11. World Health Organization (WHO): Provisional guidance on therole of specific antibiotics in the management of Mycobacte-rium ulcerans disease (Buruli ulcer). WHO, Geneva, 2006.
12. Asiedu K, Wansbrough-Jones M: Mycobacterium ulceransinfection (Buruli or Bairnsdale ulcer): challenges in developingmanagement strategies. Med J Aust 2007; 186: 55–56.
13. O’Brien DP, Hughes AJ, Cheng AC: Outcomes for Mycobacteriumulcerans infection with combined surgery and antibiotic ther-apy: findings from a south-eastern Australian case series. Med JAust 2007; 186: 58–61.
14. Chin-Lenn L, Ying D, Leong J, Ross D, Wu T, Nazaretian S, Donahoe S,Silfen R: Mycobacterium ulcerans ulcers. A proposed surgicalmanagement algorithm. Ann Plast Surg 2006; 57: 65–69.
15. Bretzel G, Siegmund V, Racz P, van Vloten F, Ngos F, ThompsonW, Biason P, Adjei O, Fleischer B, Nitschke J: Post-surgicalassessment of excised tissue from patients with Buruli ulcerdisease: progression of infection in macroscopically healthytissue. Trop Med Int Health 2005; 10: 1199–1206.
16. Rondini S, Horsfield C, Mensah-Quainoo E, Junghanss T, Lucas S,Pluschke G: Contiguous spread of Mycobacterium ulcerans inBuruli ulcer lesions analysed by histopathology and real-timePCR quantification of mycobactrial DNA. J Pathol 2006; 208:119–128.
17. Rondini S, Mensah-Quainoo E, Junghanss T, Pluschke G: Whatdoes detection of Mycobacterium ulcerans DNA in the margin ofan excised Buruli ulcer lesion tell us? J Clin Microbiol 2006; 44:4273–4275.
18. Amofah G, Asamoah S, Afram-Gyening C: Effectiveness ofexcision of pre-ulcerative Buruli lesions in field situations in arural district in Ghana. Trop Doct 1998; 28: 81–83.
20. Stinear T, Ross BC, Davies JK, Marino L, Robins-Browne RM,Oppedisano F, Sievers A, Johnson PD: Identification andcharacterization of IS2404 and IS2606: two distinct repeatedsequences for detection of Mycobacterium ulcerans by PCR.J Clin Microbiol 1999; 37: 1018–1023.
K.-H. Herbinger et al. Excision of Pre-Ulcerative Forms of Buruli Ulcer Disease
Mycobacterium ulcerans , the causative agent of buruli ulcer disease (BUD), has a wide geographical distribution in the tropical and subtropical belt, yet the clinical manifestation occurs in focal disease clusters. BUD is most common in humid, remote rural areas that are close to stagnant or slow moving bodies of water. Only in the last decade, with dramatically growing numbers of cases reported mainly from West African countries, BUD assumed importance as an emerging infectious disease. This prompted the establishment of the Global Buruli Ulcer Initiative by the World Health Organization (WHO) in 1998. 1 Presently BUD is acknowledged to be the third most frequent mycobacterial disease in humans after tuberculosis and leprosy. 2 The condition has been reported or suspected in > 30 countries worldwide. 3 West Africa is the most affected region. In Ghana, a national case search was conducted in 1999, and cases of BUD were detected in all 10 regions of the country with a peak prevalence rate of 150.8 per 100,000 in the Amansie West district. 4 Children between 2 and 15 years of age are most affected, the lesions occur predominantly on the limbs. 5 The mode of transmission is not certain, although transmission through aquatic insects has been suggested. 6,7 After inoculation of M. ulcerans into the subcutaneous tis-sue the infection initially presents as painless papule, nodule, plaque, or—rarely—as an extensive edema. Later, ulcerative skin lesions develop. The ulcers progress slowly, they are usu-ally painless and there are, besides bone involvement, no sys-temic signs of infection. 8 Although mortality is very low, the frequency of long-term sequelae is high. 9–11 Contractures and deformities causing restricted movements of affected limbs are often the result of uncontrolled self-healing processes or surgical treatment at a late, far progressed stage of disease.
Until recently wide surgical excision of BUD lesions was the only treatment option. Because there was evidence that
a combined antibiotic treatment with rifampicin and strep-tomycin has the potential to inhibit growth of M. ulcerans WHO issued provisional guidelines recommending standard antimycobacterial therapy in 2004. 12,13 Presently, prospective drug trials are being conducted comparing the efficacy of different dosages and durations of antimycobacterial treat-ment in different stages of BUD. According to the lesion category (category I, single lesion < 5 cm; category II, sin-gle lesion 5–15 cm; category III, single lesion > 15 cm and multiple lesions), different treatment schemes are applied (antibiotic treatment alone or in combination with surgery). 3 To improve the quality of clinical management, the need for follow-up studies to monitor the treatment outcome has been emphasized. 12
The purpose of this study was to assess the frequency of recurrences after surgical treatment of BUD with or without concomitant antimycobacterial treatment in two treatment centers in highly endemic areas in Ghana and to record the treatment outcome with regard to functional limitations as late sequelae of BUD.
MATERIALS AND METHODS
Study area and treatment procedures. The follow-up was conducted in the catchment area of two treatment centers for BUD in Ghana. One center, the Agogo Presbyterian Hospital, is located in the Asante Akim North District in the Ashanti Region; the other, Dunkwa Government Hospital, is situated in Dunkwa-on-Offin, Upper Denkyira District in the Central Region. BUD is highly prevalent in both areas. 4 In each treatment center one surgeon especially trained and experienced in surgical treatment of BUD performed the operations on all study subjects. For wound closure mesh skin grafting was considered the standard technique; in some cases, however, primary suturing was applied or the wound was left open for spontaneous healing. When surgery was carried out physiotherapy could not yet be offered; therefore, prevention of disability strategies were limited to providing adequate wound care and promoting early movement. Splints were not routinely applied.
* Address correspondence to Mirjam Schunk, Department of Infec-tious Diseases and Tropical Medicine (DITM), Ludwig-Maximilians University of Munich, 80802 Munich, Germany. E-mail: [email protected]
Outcome of Patients with Buruli Ulcer after Surgical Treatment with or without Antimycobacterial Treatment in Ghana
Thomas Loscher , and Gisela Bretzel Department of Infectious Diseases and Tropical Medicine (DITM), Ludwig-Maximilians University of Munich, Munich,
Germany; Agogo Presbyterian Hospital, Agogo, Ghana; Dunkwa Governmental Hospital, Dunkwa-on-Offin, Ghana; Bernhard Nocht Institute for Tropical Medicine (BNITM), Hamburg, Germany; Kumasi Centre for Collaborative Research and
Tropical Medicine (KCCR), Kwame Nkrumah University of Science and Technology Kumasi, Ghana
Abstract. This study assesses the frequency of recurrences and treatment outcome after surgery of buruli ulcer dis-ease (BUD) with or without concomitant antimycobacterial treatment. Of 129 laboratory-confirmed BUD patients who underwent surgery in two treatment centers in Ghana, 79 (61%) were retrieved for follow-up 4–29 months after the initial treatment. Among 7 (9%) recurrent cases no significant association was found between recurrences and clinical or treat-ment specific factors including antimycobacterial treatment. In 21 (27%) patients, a reduced range of motion (ROM) of one or more joints was detected. Lesions other than nodules, joint involvement, and skin grafting were identified as inde-pendent risk factors. Functional limitations hampering daily activities were perceived by 22% of the patients. Compared with other studies the recurrence rate was relatively low, functional limitations were, however, frequent. This emphasizes the need for improvement of pre- and post-treatment wound care as well as rehabilitation programs.
76 SCHUNK AND OTHERS
Patients and inclusion criteria. One hundred twenty-nine patients with a laboratory-confirmed BUD infection who received surgical treatment in the period from September 2003 to September 2005 in one of the two sites were included in the study. Laboratory confirmation was carried out by microscopy of Ziehl-Neelsen–stained smears, culture on Loewenstein-Jensen media, IS2404 polymerase chain reaction (PCR), or histopathological examination according to standardized procedures. 14,15
Data collection. Subject-specific epidemiological and clinical data was extracted from WHO “BU 01” BUD surveillance forms and hospital records, if available.
In total 16 field trips were conducted in February and March 2006: 11 in the area of Agogo and 5 in the Dunkwa area. We conducted participant interviews using a semi-structured ques-tionnaire to collect information on the pre- and post-treatment history with special focus on possible recurrences and seque-lae. For children, the guardian was interviewed.
To record the delay in seeking medical care, the time period from noticing a possible BUD lesion to presentation in a treat-ment center was documented. In addition, the site of the initial BUD lesion was examined and information on the size and the state of healing was recorded. For documentation, a pho-tograph was taken. In case of joint involvement, the passive range of motion (ROM) was measured using a standard goni-ometer (E-Z Read Goniometer 30 cm; Rolyan Jamar, Russka, Germany). ROMs were recorded by two observers according to the Neutral-Zero Measuring Method and SFTR (sagittal, frontal, and transverse rotation) documentation. 16 If any skin lesion was present, the site was clinically assessed according to WHO BUD case definition 1 and documented. In case of an ulcer, swab specimens for laboratory confirmation (micros-copy/culture/PCR) were collected. At the time of the study, assessment of punch biopsies or fine needle aspirates were not among the established methods for the routine diagnosis of BUD in Ghana; therefore, the collection of tissue specimens from non-ulcerative lesions was not possible in the field. All patients with new or not healed lesions were referred to the respective treatment center.
Definition of recurrence. In accordance to the definition of recurrences established by WHO in 2001, 1 in this study, a recurrent case was defined as a patient with previous surgical treatment with or without concomitant antimycobacterial therapy for M. ulcerans who presented with a further BUD lesion at the same or a different site within 1 year after the end of the last treatment.
Lesions that were present at follow-up were judged as BUD recurrences if the clinical features matched the BUD case definitions and the diagnosis was confirmed with at least one positive laboratory test. Clinically diagnosed BUD lesions that developed after the initial surgical excision and were sub-jected to surgery before follow-up were counted as interim recurrences. All clinically diagnosed interim recurrences were included, even if laboratory confirmation was not performed.
The time period was calculated using the date of initial sur-gery (as documented in the BU 01 form) and the date of follow-up. For interim recurrences, information from the patients was correlated with the documentation in the corresponding treat-ment center.
It must be noted that following the WHO recommenda-tion and implementation of antimycobacterial treatment, the definition of recurrence was revised by the WHO technical
advisory group. Since 2007, only new and culture confirmed BUD lesions occurring > 3 months after completion of a full course of antibiotic treatment (possibly followed by surgery), which resulted in complete healing of the initial lesion, are considered recurrences. In contrast, lesions occurring within 3 months after completion of a full course of antibiotics in the same or adjacent area are regarded ongoing cases (i.e., non-healers). In both cases, a patient is considered to have com-pleted treatment after 56 doses of antibiotics in 8 weeks. 17,18
Because treatment of our study subjects was primarily sur-gical, and none of the patients received a full course of the standardized antimycobacterial treatment according to the current WHO definitions, we consider the revised definition of recurrence not applicable.
Treatment outcome. Outcome was assessed by recording the frequency of a reduced ROM in one or more joints and the perceived functional limitations caused by BUD sequelae.
A reduced ROM was recorded when the ROM deviated from the international standardized normal range of motion. 16 A functional limitation was recorded if the patient reported to have difficulties with daily activities because of sequelae of the BU infection.
Statistical analysis. For categorical data analysis, the χ 2 test and Fisher exact test were used, and for continuous variables, Student’s t tests were applied. In addition, we constructed multiple logistic regression models for detecting potential confounders among the analyzed variables. Records with missing data for a particular variable were excluded from analysis of that variable. Calculations were done by software package SAS version 9.1. 19 P values < 0.05 were considered as statistically significant.
Ethical clearance and informed patient consent. Ethical clearance for the study was obtained through the Ethics Committee of Human Research Publication and Ethics, School of Medical Sciences, University of Science and Technology, Kumasi, Ghana. Informed consent was obtained from the participants or their guardians before beginning any study-related interventions.
RESULTS
Followed up patients. Of a total of 129 laboratory-confirmed BUD patients, 79 (61%) were retrieved for follow-up examination within a study period of 2 months. In the catchment area of Agogo, 56 (63%) of 89 patients could be followed up; in the catchment area of Dunkwa, 23 (58%) of 40 patients were retrieved. Of the 50 lost patients, 29 had a wrong address, 11 moved from the region, 3 died, and 7 could not be followed up for other reasons. The group lost to follow-up was comparable with the study participants in terms of sex, age, and year of initial treatment (data not shown).
Baseline characteristics of the followed-up patients are shown in Table 1 . Of the 79 followed-up patients, 65 (82%) received antibiotic treatment in addition to surgery as recorded in BU 01 forms and/or hospital records. In 11 of these cases, no records could be retrieved on the type and duration of treat-ment; however, in the BU 01 form, it was stated that the respec-tive patients were treated with antimycobacterial drugs. In 54 cases, hospital records with precise documentation were available: 13 patients were treated with rifampicin only, one was treated with streptomycin only, and 40 patients were treated with a combination of both. The mean duration of ther-
77OUTCOME OF PATIENTS WITH BURULI ULCER AFTER TREATMENT
apy was 39.2 ± 32.4 (SD) days for streptomycin and 26.2 ± 21.4 days for rifampicin. In 70% ( N = 38) of these patients, antimy-cobacterial treatment was started on the day of excision of the primary lesion, in 24% ( N = 13) treatment was started before surgery (minimum, 3 days; maximum, 24 days), and in 6% ( N = 3) of the cases, treatment was started after surgery (mini-mum, 2 days; maximum, 21 days). Concerning the use of drug treatment, there is less precise information of the patients in the Dunkwa region, with 43% ( N = 10) missing entries in the BU 01 forms on this topic. Records with missing data were not included in the analysis. There was no significant difference between the two treatment centers.
The time interval between the first BUD treatment and the follow-up assessment ranged between 4 and 29 months (mean, 18 months; median, 20 months). In 25% ( N = 20) of the patients, the time interval between the first BUD treat-ment and the follow-up assessment was < 12 months, in 38% ( N = 30) was 12–23 months, and in 37% ( N = 29) was > 23 months (range, 4–29 months).
Characteristics of the recurrent cases. Of the 79 patients available for follow-up, there were 11 possible recurrent cases
( Table 2 ). According to the definitions of this study, seven (9%) patients with recurrences were detected.
During the physical examination at follow-up, five patients presented with skin lesions. In four cases, small excoriations with a maximum diameter of 2 cm were located in the scar area of the previous BUD lesion. None of these lesions showed the typical clinical features of a BUD lesion, and the reported duration of the lesions was in no case > 4 weeks. The labora-tory examinations (microscopy, culture, and PCR) were nega-tive for M. ulcerans .
One patient stated at follow-up that the initial wound never fully closed after surgery 7 months earlier. On inspec-tion, we found a 2 × 13-cm purulent wound without under-mined edges on the left flank. In the Ziehl-Neelsen–stained smear of the swab sample, acid-fast bacilli were detected; cul-ture and IS2404 PCR were negative. Therefore, a superinfec-tion with other mycobacteria, as well as the possibility of a persistent, slowly healing lesion with dead bacilli, can not be excluded. In this patient, no antimycobacterial treatment had been documented.
Six patients reported interim recurrences (i.e., they had developed a recurrent lesion that was clinically diagnosed as BUD and surgically removed in the period between excision of the primary lesion and follow-up). Two of these patients came from Dunkwa and four patients were from Agogo. The median time period of the occurrence of the new lesions after excision of the primary lesion was 3 months (range, 0.5–5 months). In four cases, the patients received antimycobacterial treatment in addition to surgery for at least 14 days starting on the date of excision; in two cases, there was no written documentation of drug treatment, yet the patients remembered receiving oral and intramuscularly medication for treatment. At the time of follow-up, the patients with interim recurrences had no acute medical complaints.
In the analysis of factors possibly influencing the probabil-ity of recurrence, no significant difference was found between recurrence and size or type of lesions, use of antimycobacte-rial treatment, duration of disease before treatment, and place of treatment.
In eight patients from the follow-up group, an assessment of the margins of the excised tissue was conducted after surgical resection of the primary lesion. 20 In all of these specimens, M. ulcerans infection was detected even in the macroscopically unaffected tissue at the periphery. However, none of these patients showed a recurrence at follow-up. All had received antibiotic therapy started on the day of excision either as monotherapy with rifampicin ( N = 5) or as combination ther-apy with rifampicin and streptomycin ( N = 3) over at least 30 days.
Treatment outcome. Reduced ROM of one or more joints as a consequence of the BUD infection and treatment was detected in 21 (27%) patients. In nine of these (43%), a disabi lity caused by the primary lesion was already documented before surgical excision. The wrist was most often affected (7.6%, N = 6), followed by the joints of the hand and the ankle (6.3%, N = 5), the elbow (3.8%, N = 3), and the knee (2.5%, N = 2). Clinical features and their association to reduced ROM are summarized in Table 3 .
In 30 patients, the primary lesion was located over one or more joints, resulting in a reduced ROM of the involved joints in 60% ( N = 18) of these patients. The association between these two variables is highly significant ( P < 0.001).
Table 1 Baseline characteristics of the follow-up samples
VariablesAgogo
( N = 56)Dunkwa ( N = 23)
Total ( N = 79)
Sex (female), N (%) 28 (50) 10 (43) 38 (48)Age (years)
streptomycin/rifampicin* 38 2 40 * If documented in hospital record and/or BU 01 forms ( N = 54).
78 SCHUNK AND OTHERS
In the factor analysis, no correlation between size of the pri-mary BUD lesion and a reduced ROM was found ( P = 0.064). There was, however, a significant association between the type of primary BUD lesion and the occurrence of a reduced ROM ( P = 0.025). In contrast to other types of lesions, nodules did not result in a reduced ROM.
There was also a significant association ( P < 0.01) between the wound closure technique and the risk of a reduced ROM. From the 28 patients who received suturing after surgical inter-vention or where the lesion was left to heal spontaneously, 2 (7%) presented with a reduced ROM. Of the 51 patients where skin grafting was used, 19 (37%) showed a reduced ROM at follow-up.
From 21 patients at increased risk of a reduced range of motion (lesion other than nodule, joint involvement, skin graft for wound closure), 76% ( N = 16) presented a reduced ROM at follow-up. Multiple logistic regression models including the four above-mentioned independent variables have shown that the size of the lesion ( P = 0.71) and the type of lesion ( P = 0.93) were not significantly associated with the depen-dent variable reduced ROM. On the other hand, the involve-ment of joints ( P < 0.001) and the methods of wound closure ( P = 0.011) were significantly associated with the dependent variable. Multiple logistic regression detected that the associa-tion between the dependent variable and the size of lesion was confounded by involvement of joints. These two independent variables were highly associated ( P = 0.010), because larger lesions were significantly more often situated over joints and vice versa . Other confounding was not found.
A BUD-related functional limitation in daily activities was perceived by 22% ( N = 17) of the patients or their guard-
ians. Of the 21 patients with a reduced ROM, 52% ( N = 11) stated that they had difficulties in their daily tasks. Another six patients indicated functional restrictions, although objectively reduced ROM could not be measured.
DISCUSSION
Because there is no causative prevention of BUD, early case detection and optimization of treatment and rehabilitation are the major objectives to reduce morbidity and disability. Data on the long-term outcome after treatment are essential to evaluate treatment strategies and to adapt them according to needs identified. Despite being identified as a major public health concern in endemic areas, implementing routine after-care and follow-up is still a major problem because people affected with BUD live mostly in remote and poorly acces-sible rural areas.
The recurrence rate of 9% as detected in our study is rela-tively low compared with the results of most other published studies where recurrence rates vary from 2% to 35%. 3,21–25 However, because of variation in study design, sample sizes, follow-up periods, and diagnostic tools applied for confirma-tion of cases, the comparability of these studies is limited. In one study from Benin reporting a similar low recurrence rate of 6% ( N = 4), in total 15% ( N = 10) of the followed-up patients including two of the recurrent cases had received antimyco-bacterial treatment with rifampicin and streptomycin for up to 14 days in addition to surgical treatment. 22 A study conducted in Ghana detected a recurrence rate of 35% ( N = 27) among patients treated between 1994 and 2001 in the same two treat-ment centers as in our study. Fifty-eight percent of the patients
Table 2 Characteristics of possible recurrent cases
Sex AgeHospital
site* Inclusion†Size, type and location
of primary lesion Technique‡Type and location of
new lesion
Laboratory confirmation
of new lesion§Time
period¶Antibiotic
treatment**
Lesions at follow-upM 7 A No Medium-sized ulcer,
left kneeGrafting at second
interventionUlcer, left knee
(same site)− 8 RMP (31)
SM (19)M 14 D No Small ulcer, right arm Initial grafting Ulcer, left leg
(different site)− 29 Not
documentedF 12 D No Small ulcer, right leg Spontaneous
healingUlcer, right leg
(different site)− 28 Not
documentedM 4 D No Small plaque, left leg Spontaneous
healingUlcer, left leg
(same site)− 25 Not
documentedM 14 D Yes Medium-sized ulcer,
abdomenSpontaneous
healingUlcer, abdomen
(same site)+/− (MIC/PCR) 0 Not
documentedInterim recurrences
F 4 A Yes Medium-sized nodule, trunk
Suturing Nodule, trunk (same site)
+ (PCR) 2 RMP (37)
M 12 A Yes Medium-sized edema, right arm
Grafting at second intervention
Nodule, right arm (same site)
No testing 5 RMP (35) SM (30)
F 2 A Yes Small ulcer, right arm Grafting at second intervention
Nodule, right arm (different site)
No testing 0.5 RMP (14) SM (14)
M 3 A Yes Medium-sized ulcer, left leg
Grafting at second intervention
Nodule, right arm (different site)
− 1 RMP (30) SM (75)
M 16 D Yes Small ulcer, right arm Suturing Ulcer, left arm (different site)
+ (HIS) 3 Not documented
F 12 D Yes Medium-sized plaque, right arm
Spontaneous healing
Nodule, right arm (same site)
No testing 3 Not documented
Small < 5 cm diameter (category I); medium 5–15 cm diameter (category II); large > 15 cm diameter (category III). * A = Agogo; D = Dunkwa. † Inclusion as recurrent case. ‡ Technique of wound closure of primary lesion . § MIC = microscopy; HIS = histopathological analysis; PCR = polymerase chain reaction. ¶ Time period between excision of primary lesion and recurrence (in months). ** Antibiotic treatment with RMP (=rifampicin) and SM (=streptomycin) (duration in days).
79OUTCOME OF PATIENTS WITH BURULI ULCER AFTER TREATMENT
treated in Agogo received monotherapy with rifampicin and 9% received it in Dunkwa. 24 A prospective study from Ghana detected a recurrence rate of 20% ( N = 10) within 1 year after excision of pre-ulcerative lesions. 25 A retrospective study from the Ivory Coast examined hospital records for the frequency of BUD patients returning to the treatment center for a sec-ond intervention within 1 year and described 17% ( N = 59) recurrences. 23 In both cohorts, no antimycobacterial treatment was given.
Different factors could have contributed to the low occur-rence of recurrences in our study population. Concerning pre-disposing factors, no statistically significant influence of size and type of lesion and duration of delay of treatment seeking on the probability of recurrence was observed.
A major factor associated with the incidence of recur-rences is the expertise of surgical treatment of the primary lesion. Striking differences in healing rates between treatment centers have previously been described by other authors 24 : whereas there was a recurrence rate of 18% ( N = 6) after sur-gical treatment with wide excision margins in Agogo, there was a recurrence rate of 47% ( N = 21) in Dunkwa, where lesions were excised with no or small excision margins. In our study, however, we observed no statistically significant dif-ference concerning recurrences between the two treatment centers among patients treated between 2003 and 2005. The discrepancy between our and previously reported results may
be explained by ongoing training efforts and the implementa-tion of standardized surgical management in both treatment centers.
Another important factor contributing to the low overall recurrence rate and the diminishing difference between the two treatment centers might be the increasing use of antimy-cobacterial treatment supplementary to surgery. Recent data suggest that antibiotic treatment decreases recurrence rates to 1–4%. 13,17,18,21 Even before the WHO recommendations on antimycobacterial treatment were released, 26 both of our treat-ment centers introduced the use of a combination of strepto-mycin and rifampicin in their treatment of BUD. In total, 29% of the patient cohort treated between 1994 and 2001 in Agogo and Dunkwa received antimycobacterial drugs consisting of monotherapy with rifampicin. In contrast, 82% of patients retrieved in our follow-up study were treated with antimyco-bacterial drugs. Although the analysis of the available records showed that the treatment scheme has not yet been standard-ized, the majority of patients in both centers received a com-bination of rifampicin and streptomycin for a variable period of time.
The use of antimycobacterial treatment could also explain why there were no recurrences observed in the eight cases where M. ulcerans infection was detected in the macroscop-ically unaffected tissue of the outer margin of the excised lesions. 20 However, in our study, four of the seven patients that experienced recurrences had also received antimyco-bacterial treatment. However, considering the lack of precise data on the actual adherence to the prescribed medication and the variable duration of therapy, a positive effect of anti-mycobacterial treatment on treatment outcome can not be excluded.
Because no single factor proved to have an essential influ-ence on the occurrence of recurrent cases in the presented study, the overall reduced incidence of recurrences in this area is attributed to the multifactorial quality improvement of clini-cal management of BUD in the field of surgical as well as anti-mycobacterial treatment.
Although the low incidence of recurrences is promising, adverse sequelae (measurable ROM and/or functional limita-tions that hampered the daily activities of the patients) were frequent.
The measurement of a reduced ROM of joints can be dif-ficult in the field, and there are no validated normal ranges of motion adjusted to African populations. However, it is a use-ful tool to assess and document treatment outcomes. 27 In our study, 9 of 21 patients with reduced ROM at follow-up had presented with reduced ROM before the beginning of treat-ment. No reduced ROM was observed in cases with nodules as primary lesions, emphasizing the importance of early detec-tion and treatment. Whereas lesion over joint and lesion other than nodule and skin grafting for wound closure were identi-fied as risk factors for a reduced ROM, there was no statisti-cally significant association between occurrence of functional limitations and the size of lesion. This is a surprising result because it is obvious that large, far progressed lesions with soft tissue and bone involvement are prone to cause ankylosis and require extensive surgery, leading to significant scarring. A possible explanation for this finding is that the absolute lesion size was not related to actual body surface area. Because most patients are children, the assumed size categories do not apply equally in all patients. Also, the multiple logistic regression
* All lesions were single lesions.
Table 3 Clinical aspects and reduced range of motion (ROM)
Type and size of lesion*
Reduced ROM ( N = 21)
ROM not reduced ( N = 58) total
SuturingSkin graft
Spontaneous healing Suturing
Skin graft
Spontaneous healing
Nodule, N = 12 0 0 0 6 3 3Joint involvement:
Yes 0 0 0 1 0 0No 0 0 0 5 3 3
By sizeCategory I 0 0 0 5 3 2Category II 0 0 0 1 0 1Category III 0 0 0 0 0 0
Ulcer, N = 49 0 13 2 3 22 9Joint involvement
Yes 0 11 2 1 4 3No 0 2 0 2 18 6
By sizeCategory I 0 5 0 1 17 4Category II 0 5 2 1 5 4Category III 0 3 0 1 0 1
Edema, N = 5 0 3 0 0 2 0Joint involvement
Yes 0 3 0 0 0 0No 0 0 0 0 2 0
By sizeCategory I 0 0 0 0 0 0Category II 0 0 0 0 1 0Category III 0 3 0 0 1 0
Plaque, N = 13 0 3 0 3 5 2Joint involvement
Yes 0 2 0 2 0 1No 0 1 0 1 5 1
By sizeCategory I 0 2 0 1 0 1Category II 0 1 0 0 4 1Category III 0 0 0 2 1 0
Total 0 19 2 12 32 14
80 SCHUNK AND OTHERS
analysis indicated that the size of lesion was confounded by involvement of joints as larger lesions were significantly more often situated over joints.
Although disfigurement and joint deformities as late seque-lae of treated BUD patients are described as a common out-come, there is sparse information on its actual incidence and its social and economic impact. It has been emphasized that besides the high direct and indirect treatment costs, the impact of long-term care for disabled and handicapped BUD patients places an enormous burden on the family members. 28
The frequency of functional impairment after BUD treatment varies from 49% as observed in Ghana, 37% as reported from the Democratic Republic of Congo, to 25% as assumed by WHO. 27,29,30 The detection of a reduced ROM in a joint, however, does not necessarily correspond to the per-ceived individual functional impairment. 29 In our study, 52% ( N = 11) of the patients where a reduced ROM had been recorded stated that they had difficulties fulfilling daily tasks, whereas six patients without objectively reduced ROM indi-cated functional impairments. A BUD functional limitation score questionnaire (BUFLS) was recently developed to assess the severity of functional limitations. 9–11 After treat-ment of BUD, 57% ( N = 362) of the interviewed patients experienced one or more functional limitations according to the BULFS score. 11 In our study only 22% ( N = 17) of the patients indicated a functional impairment in daily activities. However, we did not apply the BULFS score, which may have resulted in a higher proportion of functional impairments.
The relatively high incidence of functional limitations as sequelae after BUD observed in our study sample emphasizes the need to re-evaluate the impact of different wound closure techniques on the long-term outcome and the importance of providing rehabilitative care after primary wound treatment. A main concern of follow-up activities should therefore be the establishment of programs for the prevention of disabil-ity (POD) on community level. As advised by WHO, POD activities should have a high priority in national BUD control programs.
This study had several limitations, and some conclusions should be interpreted with caution. Epidemiologic and clinical baseline data were retrieved retrospectively from BU 01 forms and hospital records. This retrospective nature of data collec-tion is prone to errors such as potential interobserver variabil-ity and assessment bias. In addition, written information on the duration and sort of drug treatment could only be obtained in 54 of 79 cases (Agogo, N = 51; Dunkwa, N = 3). Only a limited proportion of patients included in the study could be retrieved for follow-up, which might have induced a selection bias. The definition of recurrences that was valid at the time the study was conducted included all patients with a new lesion within 1 year after initial BUD treatment. Especially if the delay between primary lesion and recurrence was < 3 months and the same site was affected, a persistent rather than a recurrent lesion may have to be assumed (non-healer). At the time when our study was conducted, the main focus of the treatment cen-ters was the identification of new cases. Therefore, laboratory confirmation of interim recurrences was not performed on a regular basis. Of the few cases subjected to laboratory analy-sis, unfortunately, positive culture results were not obtained. As reflected in the revised WHO definition of recurrences, the presence of acid-fast bacilli and/or a positive PCR signal for IS2404 might equally well reflect the presence of dead
bacilli. 17,18 The diagnosis of recurrence in our study patients was always established by clinicians with long-standing experi-ence in BUD. However, we concede that in view of the revised WHO definition, the recurrent cases identified in this study may not fully meet the criteria and may rather be referred to as clinically suspect recurrences.
Received October 30, 2008. Accepted for publication April 15, 2009.
Authors’ addresses : Mirjam Schunk, Erna Fleischmann, Vera Siegmund, Karl-Heinz Herbinger, Thomas Loscher, and Gisela Bretzel, Depar t-ment of Infectious Diseases and Tropical Medicine (DITM), Ludwig-Maximilians University of Munich, Leopoldstrasse 5, 80802 Munich, Germany. William Thompson, Agogo Presbyterian Hospital, Agogo, Ghana. Erasmus Klutse, Dunkwa Governmental Hospital, Dunkwa- on-Offin, Ghana. Jörg Nitschke and Bernhard Fleischer, Bernhard Nocht Institute for Tropical Medicine (BNITM), Bernhard-Nocht Strasse 74, 20359 Hamburg, Germany. Kwame Opare-Asamoah, Ruth Thompson, and Ohene Adjei, Kumasi Center for Collaborative Research in Tropical Medicine (KCCR), Kwame Nkrumah University of Science and Technology (KNUST), University Post Office, Kumasi, Ghana.
REFERENCES
1. World Health Organization, 2001. Buruli ulcer. Management of Mycobacterium ulcerans disease. Available at: http://www.whqlibdoc.who.int/hq/2001/WHO_CDS_CPE_GBU I_2001.3.pdf. Accessed August 26, 2008.
2. van der Werf TS, van der Graaf WT, Tappero JW, Asiedu K, 1999. Mycobacterium ulcerans infection. Lancet 354: 1013–1018.
3. World Health Organization, 2008. Buruli ulcer: progress report, 2004–2008. Wkly Epidemiol Rec 17: 145–156.
4. Amofah G, Bonsu F, Tetteh C, Okrah J, Asamoa K, Asiedu K, Addy J, 2002. Buruli ulcer in Ghana: results of a national case search. Emerg Infect Dis 8: 167–170.
5. Wansbrough-Jones M, Phillips R, 2006. Buruli ulcer: emerging from obscurity. Lancet 367: 1849–1858.
6. Portaels F, Elsen P, Guimaraes-Peres A, Fonteyne PA, Meyers WM, 1999. Insects in the transmission of Mycobacterium ulcer-ans infection. Lancet 353: 986.
7. Portaels F, Meyers WM, Ablordey A, Castro AG, 2008. First culti-vation and characterization of Mycobacterium ulcerans from the environment. PLoS Negl Trop Dis 2: e178.
9. Stienstra Y, Dijkstra PU, Guédénon A, Johnson RC, Ampadu EO, Mensah T, Klutse EY, Etuaful S, Deepak S, van der Graaf WT, van der Werf TS, 2004. Development of a questionnaire assess-ing Buruli ulcer-induced functional limitation. Am J Trop Med Hyg 70: 318–322.
10. Stienstra Y, Dijkstra PU, van Wezel MJ, van Roest MH, Beets M, Zijlstra I, Johnson RC, Ampadu EO, Gbovi J, Zinsou C, Etuaful S, Klutse EY, Van der Graaf WT, Van der Werf TS, 2005. Reliability and validity of the buruli ulcer functional limitation score questionnaire. Am J Trop Med Hyg 72: 449–452.
11. Stienstra Y, van Roest MHG, van Wezel MJ, Wiersma IC, Hospers IC, Dijkstra PU, Johnson RC, Ampadu EO, Gbovi J, Zinsou C, Etuaful S, Klutse EY, van der Graaf WT, van der Werf TS, 2005. Factors associated with functional limitations and subsequent employment or schooling in Buruli ulcer patients. Trop Med Int Health 10: 1251–1257.
12. World Health Organization, 2004. Provisional guidance on the role of antibiotics in the management of Mycobacterium ulcer-ans disease. Available at: http://www.who.int /entity/buruli/ information/antibiotics/en/. Accessed August 26, 2008.
13. Etuaful S, Carbonnelle B, Grosset J, Lucas S, Horsfield C, Phillips R, Evans M, Ofori-Adjei D, Klustse E, Owusu-Boateng J, Amedofu GK, Awuah P, Ampadu E, Amofah G, Asiedu K, Wansbrough-Jones M, 2005. Efficacy of the combination rifampin-streptomycin in preventing growth of Mycobacterium ulcerans in early lesions of Buruli ulcer in humans. Antimicrob Agents Chemother 49: 3182–3186.
81OUTCOME OF PATIENTS WITH BURULI ULCER AFTER TREATMENT
14. World Health Organization, 2001. Buruli ulcer: diagnosis of Myco bacterium ulcerans disease. Available at: http://whqlibdoc.who.int/hq/2001/WHO_CDS_CPE_GBUI_2001.4.pdf. Accessed August 26, 2008.
15. Bretzel G, Siegmund V, Nitschke J, Herbinger KH, Thompson W, Klutse E, Crofts K, Massavon W, Etuaful S, Thompson R, Asamoah-Opare K, Racz P, Vloten F, van Berberich C, Kruppa T, Ampadu E, Fleischer B, Adjei O, 2007. A step-wise approach to the laboratory diagnosis of Buruli ulcer disease. Trop Med Int Health 12: 89–96.
16. Gerhardt JJ, Rondinelli RD, 2001. Goniometric techniques for range-of-motion assessment. Phys Med Rehabil Clin N Am 12: 507–527.
17. World Health Organization, 2007. Summary report of the WHO annual meeting on Buruli Ulcer, 2–4 April 2007 and report of the technical advisory Group (TAG) meeting, 5 April, Geneva, 2007. Available at: http://www.who.int/buruli/events/Report_2007_meeting_FINAL.pdf. Accessed August 26, 2008.
18. World Health Organization, 2008. Meeting of the WHO technical advisory group on Buruli ulcer 3 April 2008, Geneva, summary report. Available at: http://www.who.int/buruli/events/TAG_report_FINAL.pdf. Accessed August 26, 2008.
19. SAS Institute, 2006. SAS 9.1.3. SAS Institute, Cary, NC. 20. Bretzel G, Siegmund V, Racz P, van Vloten F, Ngos F, Thompson W,
Biason P, Adjei O, Fleischer B, Nitschke J, 2005. Post-surgical assessment of excised tissue from patients with Buruli ulcer dis-ease: progression of infection in macroscopically healthy tissue. Trop Med Int Health 10: 1199–1206.
21. Chauty A, Ardant MF, Adeye A, Euverte H, Guédénon A, Johnson C, Aubry J, Nuermberger E, Grosset J, 2007. Promising clinical efficacy of streptomycin-rifampin combination for treatment of buruli ulcer ( Mycobacterium ulcerans disease). Antimicrob Agents 51: 4029–4035.
23. Kanga JM, Kacou DE, Sangaré A, Dabila Y, Asse NH, Djakeaux S, 2003. [Recurrence cases observed after surgical treatment of Buruli ulcer in Cote d’Ivoire]. Bull Soc Pathol Exot 96: 406–409.
24. Teelken MA, Stienstra Y, Ellen DE, Quarshie E, Klutse E, van der Graaf WT, van der Werf TS, 2003. Buruli ulcer: differences in treatment outcome between two centres in Ghana. Acta Trop 88: 51–56.
25. Amofah G, Asamoah S, Afram-Gyening C, 1998. Effectiveness of excision of preulcerative Buruli lesions in field situations in a rural district in Ghana. Trop Doct 28: 81–83.
26. World Health Organization, 2004. Buruli ulcer disease. Wkly Epidemiol Rec 79: 145–149.
27. World Health Organization, 2006. Buruli ulcer: a manual on how to prevent disability. Available at: http://www.who.int/buruli/information/publications/pod/en/index.html. Accessed August 26, 2008.
28. Asiedu K, Etuaful S, 1998. Socioeconomic implications of Buruli ulcer in Ghana: a three-year review. Am J Trop Med Hyg 59: 1015–1022.
29. Ellen DE, Stienstra Y, Teelken MA, Dijkstra PU, van der Graaf WT, van der Werf TS, 2003. Assessment of functional limita-tions caused by Mycobacterium ulcerans infection: towards a Buruli ulcer functional limitation score. Trop Med Int Health 8: 90–96.
30. Phanzu DM, Bafende EA, Dunda BK, Imposo DB, Kibadi AK, Nsiangana SZ, Singa JN, Meyers WM, Suykerbuyk P, Portaels F, 2006. Mycobacterium ulcerans disease (Buruli ulcer) in a rural hospital in Bas-Congo, Democratic Republic of Congo, 2002–2004. Am J Trop Med Hyg 75: 311–314.